Search

Background/Objectives: Radial artery tracking (RAT) in the short-axis view is a pivotal step for ultrasound-guided radial artery catheterization (RAC), which is widely employed in various clinical settings. To eliminate disparities and lay the foundations for automated procedures, a pilot study was conducted to explore the feasibility of U-Net and its variants in automatic RAT. Methods: Approved by the institutional ethics committee, patients as potential RAC candidates were enrolled, and the radial arteries were continuously scanned by B-mode ultrasonography. All acquired videos were processed into standardized images, and randomly divided into training, validation, and test sets in an 8:1:1 ratio. Deep learning models, including U-Net and its variants, such as Attention U-Net, UNet++, Res-UNet, TransUNet, and UNeXt, were utilized for automatic RAT. The performance of the deep learning architectures was assessed using loss functions, dice similarity coefficient (DSC), and Jaccard similarity coefficient (JSC). Performance differences were analyzed using the Kruskal–Wallis test. Results: The independent datasets comprised 7233 images extracted from 178 videos of 135 patients (53.3% women; mean age: 41.6 years). Consistent convergence of loss functions between the training and validation sets was achieved for all models except Attention U-Net. Res-UNet emerged as the optimal architecture in terms of DSC and JSC (93.14% and 87.93%), indicating a significant improvement compared to U-Net (91.79% vs. 86.19%, p < 0.05) and Attention U-Net (91.20% vs. 85.02%, p < 0.05). Conclusions: This pilot study validates the feasibility of U-Net and its variants in automatic RAT, highlighting the predominant performance of Res-UNet among the evaluated architectures. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (DCNNs) have attained human-level performance for object categorization and exhibited representation alignment between network layers and brain regions. Does such representation alignment naturally extend to other visual tasks beyond recognizing objects in static images? In this study, we expanded the exploration to the recognition of human actions from videos and assessed the representation capabilities and alignment of two-stream DCNNs in comparison with brain regions situated along ventral and dorsal pathways. Using decoding analysis and representational similarity analysis, we show that DCNN models do not show hierarchical representation alignment to human brain across visual regions when processing action videos. Instead, later layers of DCNN models demonstrate greater representation similarities to the human visual cortex. These findings were revealed for two display formats: photorealistic avatars with full-body information and simplified stimuli in the point-light display. The discrepancies in representation alignment suggest fundamental differences in how DCNNs and the human brain represent dynamic visual information related to actions. [ABSTRACT FROM AUTHOR]

The Kalman filter and extended Kalman filter algorithm are widely used in physical education video processing, but their performance still needs to be improved for complex backgrounds or high dynamic targets. This paper presents an interactive multi-model algorithm. The displacement detection Kalman filter is used to track moving objects. This method can solve the problem of a single model not being able to match the motion features well. Finally, the simulation experiment of football match video proves that the proposed method can significantly improve the tracking accuracy of moving objects in video. [ABSTRACT FROM AUTHOR]

Emotion recognition in user‐generated videos plays an essential role in affective computing. In general, visual information directly affects human emotions, so the visual modality is significant for emotion recognition. Most classic approaches mainly focus on local temporal information of videos, which potentially restricts their capacity to encode the correlation of long‐range context. To address this issue, a novel network is proposed to recognize emotions in videos. To be specific, a spatio‐temporal correlation‐aware block is designed to depict the long‐range correlations between input tokens, where the convolutional layers are used to learn the local correlations and the inter‐image cross‐attention is designed to learn the long‐range and spatio‐temporal correlations between input tokens. To generate diverse and challenging samples, a dual‐augmentation fusion layer is devised, which fuses each frame with its corresponding frame in the temporal domain. To produce rich video clips, a long‐range sampling layer is designed, which generates clips in a wide range of spatial and temporal domains. Extensive experiments are conducted on two challenging video emotion datasets, namely VideoEmotion‐8 and Ekman‐6. The experimental results demonstrate that the proposed method obtains better performance than baseline methods. Moreover, the proposed method achieves state‐of‐the‐art results on the two datasets. The source code of the proposed network is available at: https://github.com/JinChow/LRCANet. [ABSTRACT FROM AUTHOR]

As internet speeds increase and user demands for video quality grow, video coding standards continue to evolve. H.266/Versatile Video Coding (VVC), as the new generation of video coding standards, further improves compression efficiency but also brings higher computational complexity. Despite the significant advancements VVC has made in compression ratio and video quality, the introduction of new coding techniques and complex coding unit (CU) partitioning methods have also led to increased encoding complexity. This complexity not only extends encoding time but also increases hardware resource consumption, limiting the application of VVC in real-time video processing and low-power devices.To alleviate the encoding complexity of VVC, this paper puts forward a Bayesian and texture-feature-based fast splitting algorithm for coding intraframe bloc of VVC, which aims to reduce unnecessary computational steps, enhance encoding efficiency, and maintain video quality as much as possible. In the stage of rapid coding, the video frames are coded by the original VVC test model (VTM), and Joint Rough Mode Decision (JRMD) evaluation cost is used to update the parameter in the Bayesian algorithm to come and set the two thresholds to judge whether the current coding block continues to be split or not. Then, for coding blocks larger than those satisfying the above threshold conditions, the predominant direction of the texture within the coding block is ascertained by calculating the standard deviations along both the horizontal and vertical axes so as to skip some unnecessary splits in the current coding block patterns. The findings from our experiments demonstrate that our proposed approach improves the encoding rate by 1.40% on average, and the execution time of the encoder has been reduced by 49.50%. The overall algorithm has optimized the VVC intraframe coding technology and reduced the coding complexity of VVC. [ABSTRACT FROM AUTHOR]

In this paper, we propose an efficient target detection algorithm, ASF-Sim-YOLO, to address issues encountered in escalator step defect detection, such as an excessive number of parameters in the detection network model, poor adaptability, and difficulties in real-time processing of video streams. Firstly, to address the characteristics of escalator step defects, we designed the ASF-Sim-P2 structure to improve the detection accuracy of small targets, such as step defects. Additionally, we incorporated the SimAM (Similarity-based Attention Mechanism) by combining SimAM with SPPF (Spatial Pyramid Pooling-Fast) to enhance the model's ability to capture key information by assigning importance weights to each pixel. Furthermore, to address the challenge posed by the small size of step defects, we replaced the traditional CIoU (Complete-Intersection-over-Union) loss function with NWD (Normalized Wasserstein Distance), which alleviated the problem of defect missing. Finally, to meet the deployment requirements of mobile devices, we performed channel pruning on the model. The experimental results showed that the improved ASF-Sim-YOLO model achieved an average accuracy (mAP50) of 96.8% on the test data set, which was a 22.1% improvement in accuracy compared to the baseline model. Meanwhile, the computational complexity (in GFLOPS) of the model was reduced to a quarter of that of the baseline model, while the frame rate (FPS) was improved to 575.1. Compared with YOLOv3-tiny, YOLOv5s, YOLOv8s, Faster-RCNN, TOOD, RTMDET and other deep learning-based target recognition algorithms, ASF-Sim-YOLO has better detection accuracy and real-time processing capability. These results demonstrate that ASF-Sim-YOLO effectively balances lightweight design and performance improvement, making it highly suitable for real-time detection of step defects, which can meet the demands of escalator inspection operations. [ABSTRACT FROM AUTHOR]

As the performance of machine vision continues to improve, it is being used in various industrial fields to analyze and generate massive amounts of video data. Although the demand for and consumption of video data by machines has increased significantly, video coding for machines needs to be improved. It is therefore necessary to consider a new codec that differs from conventional codecs based on the human visual system (HVS). Spatial down-sampling plays a critical role in video coding for machines because it reduces the volume of the video data to be processed while maintaining the shape of the data's features that are important for the machine to reference when processing the video. An effective method of determining the intensity of spatial down-sampling as an efficient coding tool for machines is still in the early stages. Here, we propose a method of determining an optimal scale factor for spatial down-sampling by collecting and analyzing information on the number of objects and the ratio of the area occupied by the object within a picture. We compare the data reduction ratio to the machine accuracy error ratio (DRAER) to evaluate the performance of the proposed method. By applying the proposed method, the DRAER was found to be a maximum of 21.40 dB and a minimum of 11.94 dB. This shows that video coding gain for the machines could be achieved through the proposed method while maintaining the accuracy of machine vision tasks. [ABSTRACT FROM AUTHOR]

Human action understanding (HAU) is a broad topic that involves specific tasks, such as action localisation, recognition, and assessment. However, most popular HAU datasets are bound to one task based on particular actions. Combining different but relevant HAU tasks to establish a unified action understanding system is challenging due to the disparate actions across datasets. A large‐scale and comprehensive benchmark, namely SkatingVerse is constructed for action recognition, segmentation, proposal, and assessment. SkatingVerse focus on fine‐grained sport action, hence figure skating is chosen as the task object, which eliminates the biases of the object, scene, and space that exist in most previous datasets. In addition, skating actions have inherent complexity and similarity, which is an enormous challenge for current algorithms. A total of 1687 official figure skating competition videos was collected with a total of 184.4 h, exceeding four times over other datasets with a similar topic. SkatingVerse enables to formulate a unified task to output fine‐grained human action classification and assessment results from a raw figure skating competition video. In addition, SkatingVerse can facilitate the study of HAU foundation model due to its large scale and abundant categories. Moreover, image modality is incorporated for human pose estimation task into SkatingVerse. Extensive experimental results show that (1) SkatingVerse significantly helps the training and evaluation of HAU methods, (2) the performance of existing HAU methods has much room to improve, and SkatingVerse helps to reduce such gaps, and (3) unifying relevant tasks in HAU through a uniform dataset can facilitate more practical applications. SkatingVerse will be publicly available to facilitate further studies on relevant problems. [ABSTRACT FROM AUTHOR]

Accurate assessment of water surface velocity (WSV) is essential for flood prevention, disaster mitigation, and erosion control within hydrological monitoring. Existing image-based velocimetry techniques largely depend on correlation principles, requiring users to input and adjust parameters to achieve reliable results, which poses challenges for users lacking relevant expertise. This study presents RivVideoFlow, a user-friendly, rapid, and precise method for WSV. RivVideoFlow combines two-dimensional and three-dimensional orthorectification based on Ground Control Points (GCPs) with a deep learning-based multi-frame optical flow estimation algorithm named VideoFlow, which integrates temporal cues. The orthorectification process employs a homography matrix to convert images from various angles into a top-down view, aligning the image coordinates with actual geographical coordinates. VideoFlow achieves superior accuracy and strong dataset generalization compared to two-frame RAFT models due to its more effective capture of flow velocity continuity over time, leading to enhanced stability in velocity measurements. The algorithm has been validated on a flood simulation experimental platform, in outdoor settings, and with synthetic river videos. Results demonstrate that RivVideoFlow can robustly estimate surface velocity under various camera perspectives, enabling continuous real-time dynamic measurement of the entire flow field. Moreover, RivVideoFlow has demonstrated superior performance in low, medium, and high flow velocity scenarios, especially in high-velocity conditions where it achieves high measurement precision. This method provides a more effective solution for hydrological monitoring. [ABSTRACT FROM AUTHOR]

In order to solve the problem of low accuracy and efficiency in video human behavior recognition algorithm, a deep learning video human behavior recognition algorithm is proposed, which is based on an improved time division network. This method innovates on the classical two-stream convolutional neural network framework, and the core is to enhance the performance of the time division network by implementing the sliding window sampling technique with multiple time scales. This sampling strategy not only effectively integrates the full time-series information of the video, but also accurately captures the long-term dependencies hidden in human behavior, which further improves the accuracy and efficiency of behavior recognition. Experimental results show that the method proposed in this paper has achieved good advantages in multiple data sets. On HMDB51, our method achieves 84% recognition accuracy, while on the more complex Kinetics and UCF101 datasets, it also achieves 94% and significant recognition results, respectively. In the face of complex scenes and changeable human body structure, the proposed algorithm shows excellent robustness and stability. In terms of real-time, it can meet the high requirements of real-time video processing. Through the validation of experimental data, our method has made significant progress in extracting spatiotemporal features, capturing long-term dependencies, and focusing on key information. [ABSTRACT FROM AUTHOR]

In order to reduce the encoding complexity and stream size, improve the encoding performance and further improve the compression performance, the depth prediction partition encoding is studied in this paper. In terms of pattern selection strategy, optimization analysis is carried out based on fast strategic decision-making methods to ensure the comprehensiveness of data processing. In the design of adaptive strategies, different adaptive quantization parameter adjustment strategies are adopted for the equatorial and polar regions by considering the different levels of user attention in 360 degree virtual reality videos. The purpose is to achieve the optimal balance between distortion and stream size, thereby managing the output stream size while maintaining video quality. The results showed that this strategy achieved a maximum reduction of 2.92% in bit rate and an average reduction of 1.76%. The average coding time could be saved by 39.28%, and the average reconstruction quality was 0.043, with almost no quality loss detected by the audience. At the same time, the model demonstrated excellent performance in sequences of 4K, 6K, and 8K. The proposed deep partitioning adaptive strategy has significant improvements in video encoding quality and efficiency, which can improve encoding efficiency while ensuring video quality. [ABSTRACT FROM AUTHOR]

Existing lightweight video super-resolution network architectures are often simple in structure and lack generalization ability when dealing with complex and varied real scenes in aerial videos of unmanned aerial vehicle. Furthermore, these networks may cause issues such as the checkerboard effect and loss of texture information when processing drone videos. To address these challenges, we propose a super-lightweight video super-resolution reconstruction network based on convolutional pyramids and progressive residual blocks: ULSR-UV. The ULSR-UV network significantly reduces model redundancy and achieves high levels of lightness by incorporating a 3D lightweight spatial pyramid structure and more efficient residual block designs. This network utilizes a specific optimizer to efficiently process drone videos from both multi-frame and single-frame dimensions. Additionally, the ULSR-UV network incorporates a multidimensional feature loss calculation module that enhances network performance and significantly improves the reconstruction quality of drone aerial videos. Extensive experimental verification has demonstrated ULSR-UV's outstanding performance in the field of drone video super-resolution reconstruction. [ABSTRACT FROM AUTHOR]

Urban public safety management relies heavily on video surveillance systems, which provide crucial visual data for resolving a wide range of incidents and controlling unlawful activities. Traditional methods for target detection predominantly employ a two-stage approach, focusing on precision in identifying objects such as pedestrians and vehicles. These objects, typically sparse in large-scale, lower-quality surveillance footage, induce considerable redundant computation during the initial processing stage. This redundancy constrains real-time detection capabilities and escalates processing costs. Furthermore, transmitting raw images and videos laden with superfluous information to centralized back-end systems significantly burdens network communications and fails to capitalize on the computational resources available at diverse surveillance nodes. This study introduces DiffRank, a novel preprocessing method for fixed-angle video imagery in urban surveillance. The method strategically generates candidate regions during preprocessing, thereby reducing redundant object detection and improving the efficiency of the detection algorithm. Drawing upon change detection principles, a background feature learning approach utilizing shallow features has been developed. This approach prioritizes learning the characteristics of fixed-area backgrounds over direct background identification. As a result, alterations in ROI are efficiently discerned using computationally efficient shallow features, markedly accelerating the generation of proposed Regions of Interest (ROIs) and diminishing the computational demands for subsequent object detection and classification. Comparative analysis on various public and private datasets illustrates that DiffRank, while maintaining high accuracy, substantially outperforms existing baselines in terms of speed, particularly with larger image sizes (e.g., an improvement exceeding 300 % at 1920×1080 resolution). Moreover, the method demonstrates enhanced robustness compared to baseline methods, efficiently disregarding static targets like mannequins in display windows. The advancements in candidate area preprocessing enable a balanced approach between detection accuracy and overall detection speed, making the algorithm highly applicable for real-time on-site analysis in edge computing scenarios and cloud-edge collaborative computing environments. [ABSTRACT FROM AUTHOR]

The United States is predicted to experience an annual decline in milk production due to heat stress of 1.4 and 1.9 kg/day by the 2050s and 2080s, with economic losses of USD 1.7 billion and USD 2.2 billion, respectively, despite current cooling efforts implemented by the dairy industry. The ability of cattle to withstand heat (i.e., thermotolerance) can be influenced by physiological and behavioral factors, even though the factors contributing to thermoregulation are heritable, and cows vary in their behavioral repertoire. The current methods to gauge cow behaviors are lacking in precision and scalability. This paper presents an approach leveraging various machine learning (ML) (e.g., CNN and YOLOv8) and computer vision (e.g., Video Processing and Annotation) techniques aimed at quantifying key behavioral indicators, specifically drinking frequency and brush use- behaviors. These behaviors, while challenging to quantify using traditional methods, offer profound insights into the autonomic nervous system function and an individual cow's coping mechanisms under heat stress. The developed approach provides an opportunity to quantify these difficult-to-measure drinking and brush use behaviors of dairy cows milked in a robotic milking system. This approach will open up a better opportunity for ranchers to make informed decisions that could mitigate the adverse effects of heat stress. It will also expedite data collection regarding dairy cow behavioral phenotypes. Finally, the developed system is evaluated using different performance metrics, including classification accuracy. It is found that the YoloV8 and CNN models achieved a classification accuracy of 93% and 96% for object detection and classification, respectively. [ABSTRACT FROM AUTHOR]

The integration of sign language recognition systems into digital platforms has the potential to bridge communication gaps between the deaf community and the broader population. This paper introduces an advanced Israeli Sign Language (ISL) recognition system designed to interpret dynamic motion gestures, addressing a critical need for more sophisticated and fluid communication tools. Unlike conventional systems that focus solely on static signs, our approach incorporates both deep learning and Computer Vision techniques to analyze and translate dynamic gestures captured in real-time video. We provide a comprehensive account of our preprocessing pipeline, detailing every stage from video collection to the extraction of landmarks using MediaPipe, including the mathematical equations used for preprocessing these landmarks and the final recognition process. The dataset utilized for training our model is unique in its comprehensiveness and is publicly accessible, enhancing the reproducibility and expansion of future research. The deployment of our model on a publicly accessible website allows users to engage with ISL interactively, facilitating both learning and practice. We discuss the development process, the challenges overcome, and the anticipated societal impact of our system in promoting greater inclusivity and understanding. [ABSTRACT FROM AUTHOR]

The integration of sign language recognition systems into digital platforms has the potential to bridge communication gaps between the deaf community and the broader population. This paper introduces an advanced Israeli Sign Language (ISL) recognition system designed to interpret dynamic motion gestures, addressing a critical need for more sophisticated and fluid communication tools. Unlike conventional systems that focus solely on static signs, our approach incorporates both deep learning and Computer Vision techniques to analyze and translate dynamic gestures captured in real-time video. We provide a comprehensive account of our preprocessing pipeline, detailing every stage from video collection to the extraction of landmarks using MediaPipe, including the mathematical equations used for preprocessing these landmarks and the final recognition process. The dataset utilized for training our model is unique in its comprehensiveness and is publicly accessible, enhancing the reproducibility and expansion of future research. The deployment of our model on a publicly accessible website allows users to engage with ISL interactively, facilitating both learning and practice. We discuss the development process, the challenges overcome, and the anticipated societal impact of our system in promoting greater inclusivity and understanding.

Markerless motion capture has been made available by the development of pose estimation algorithms that provide both an estimate of the location of body keypoints in two-dimensional images and its associated confidence. It seems relevant to use this additional information for three-dimensional (3D) point reconstruction. Yet, it has been little described, nor has its influence on 3D point reconstruction. Eight participants performed a manual material handling task, which was recorded by 10 video cameras. Each video was processed using OpenPose. Different 3D point reconstruction methods were compared: direct linear transform (DLT) and weighted DLT (wDLT), with 10 cameras and with a subset of 4 cameras. For each keypoint, confidence and position deviation from the 3D reconstructed point with 10 cameras projected in the image reference frame were assessed. Results suggest that using confidence information reduces both average and maximum 3D distance between 3D points reconstructed with 4 and 10 cameras. [ABSTRACT FROM AUTHOR]

Most previous few-shot action recognition works tend to process video temporal and spatial features separately, resulting in insufficient extraction of comprehensive features. In this paper, a novel hybrid attentive prototypical network (HAPN) framework for few-shot action recognition is proposed. Distinguished by its joint processing of temporal and spatial information, the HAPN framework strategically manipulates these dimensions from feature extraction to the attention module, consequently enhancing its ability to perform action recognition tasks. Our framework utilizes the R(2+1)D backbone network, coupling the extraction of integrated temporal and spatial features to ensure a comprehensive understanding of video content. Additionally, our framework introduces the novel Residual Tri-dimensional Attention (ResTriDA) mechanism, specifically designed to augment feature information across the temporal, spatial, and channel dimensions. ResTriDA dynamically enhances crucial aspects of video features by amplifying significant channel-wise features for action distinction, accentuating spatial details vital for capturing the essence of actions within frames, and emphasizing temporal dynamics to capture movement over time. We further propose a prototypical attentive matching module (PAM) built on the concept of metric learning to resolve the overfitting issue common in few-shot tasks. We evaluate our HAPN framework on three classical few-shot action recognition datasets: Kinetics-100, UCF101, and HMDB51. The results indicate that our framework significantly outperformed state-of-the-art methods. Notably, the 1-shot task, demonstrated an increase of 9.8% in accuracy on UCF101 and improvements of 3.9% on HMDB51 and 12.4% on Kinetics-100. These gains confirm the robustness and effectiveness of our approach in leveraging limited data for precise action recognition. [ABSTRACT FROM AUTHOR]

Violence can be committed anywhere, even in crowded places. It is hence necessary to monitor human activities for public safety. Surveillance cameras can monitor surrounding activities but require human assistance to continuously monitor every incident. Automatic violence detection is needed for early warning and fast response. However, such automation is still challenging because of low video resolution and blind spots. This paper uses ResNet50v2 and the gated recurrent unit (GRU) algorithm to detect violence in the Movies, Hockey, and Crowd video datasets. Spatial features were extracted from each frame sequence of the video using a pretrained model from ResNet50V2, which was then classified using the optimal trained model on the GRU architecture. The experimental results were then compared with wavelet feature extraction methods and classification models, such as the convolutional neural network and long short-term memory. The results show that the proposed combination of ResNet50V2 and GRU is robust and delivers the best performance in terms of accuracy, recall, precision, and F1-score. The use of ResNet50V2 for feature extraction can improve model performance.

Video Anomaly Detection (VAD) has been an active research field for several decades. However, most existing approaches merely extract a single type of feature from videos and define a single paradigm to indicate the extent of abnormalities. A coarse‐to‐fine three‐level prediction is built by integrating different levels of spatio‐temporal representations, better highlighting the difference between normal and abnormal behaviors. First, an object‐level trajectory prediction is proposed to model human historical position using a graph transformer network. Subsequently, skeleton‐level prediction is achieved by incorporating the positional information from the trajectory prediction. More importantly, based on the predicted skeleton, a skeleton‐guided pixel‐level region prediction is performed. A novel Skeleton Conditioned Generative Adversarial Network (SCGAN) is designed to explore the correlation between skeleton‐level and pixel‐level motion prediction. Benefiting from SCGAN, the prediction of human regions is contributed by both coarse‐grained and fine‐grained motion features. This three‐level prediction, namely Progressive Prediction Video Anomaly Detection (P3VAD), enlarges the prediction error on irregular motion patterns. Besides, a pixel‐level analysis method is proposed to achieve Background‐bias Elimination (BE) and denoise the predicted region. Experimental results validate the effectiveness of P3VAD on the four benchmark datasets (ShanghaiTech, CUHK Avenue, IITB‐Corridor, and ADOC). [ABSTRACT FROM AUTHOR]

Speech disorders are significant barriers to the balanced development of a child. Many children in Poland are affected by lisps (sigmatism)—the incorrect articulation of sibilants. Since speech therapy diagnostics is complex and multifaceted, developing computer-assisted methods is crucial. This paper presents the results of assessing the usefulness of hybrid feature vectors extracted based on multimodal (video and audio) data for the place of articulation assessment in sibilants /s/ and /ʂ/. We used acoustic features and, new in this field, visual parameters describing selected articulators' texture and shape. Analysis using statistical tests indicated the differences between various sibilant realizations in the context of the articulation pattern assessment using hybrid feature vectors. In sound /s/, 35 variables differentiated dental and interdental pronunciation, and 24 were visual (textural and shape). For sibilant /ʂ/, we found 49 statistically significant variables whose distributions differed between speaker groups (alveolar, dental, and postalveolar articulation), and the dominant feature type was noise-band acoustic. Our study suggests hybridizing the acoustic description with video processing provides richer diagnostic information. [ABSTRACT FROM AUTHOR]

One-photon fluorescent calcium imaging helps understand brain functions by recording large-scale neural activities in freely moving animals. Automatic, fast, and accurate active neuron segmentation algorithms are essential to extract and interpret information from these videos. One-photon imaging videos' low resolution, high noise, and high background fluctuation pose significant challenges. Here, we develop a software pipeline to address the challenges of processing one-photon calcium imaging videos. We extend our previous two-photon active neuron segmentation algorithm, Shallow U-Net Neuron Segmentation (SUNS), to better suppress background fluctuations in one-photon videos. We also develop additional neuron extraction (ANE) to locate small or dim neurons missed by SUNS. To train our segmentation method, we create ground truth neurons by developing a manual labeling pipeline assisted with semi-automatic refinement. Our method is more accurate and faster than state-of-the-art techniques when processing simulated videos and multiple experimental datasets acquired over various brain regions with different imaging conditions. SUNS2-ANE segments active neurons from one-photon calcium imaging videos accurately and fast. It improves background suppression and extracts additional weak neurons. A separate semi-automatic labeling pipeline generates ground truth for training. [ABSTRACT FROM AUTHOR]

The purpose of the current study was to develop a database of gaming photo stimuli to be used in future psychological research assessing behavioral, cognitive, and neural correlates related to gaming. Participants (ages 18-42, N = 549; 43.17% male) completed ratings on 119 gaming-related images across 5 different categories: valence, arousal, relevance, urge, and interest. A measure of gaming addiction was also included. Positive associations between gaming addiction scores and image ratings were predicted. Gamers rated images higher than non-gamers across multiple dimensions including valence (p = .0012), arousal (p < .0001), urge (p < .0001), and interest (p < .0001). Gaming addiction scores were positively associated with image ratings for valence, r = .399, arousal, r = .438, relevance, r = .215, urge, r = .550, and interest, r = .523, p < .0001. Finally, average image ratings for the overall sample ranged from 5.65 (SD = 2.04) to 3.63 (SD = 1.91) for relevance and interest, respectively. These findings suggest that databases of video gaming imagery, rated for valence, arousal, relevance, urge, and interest, could possibly be used in studies assessing cognitive processing of video gaming-related stimuli in individuals with problematic gaming behavior. [ABSTRACT FROM AUTHOR]

This article introduces Lester, a novel method to automatically synthesize retro-style 2D animations from videos. The method approaches the challenge mainly as an object segmentation and tracking problem. Video frames are processed with the Segment Anything Model (SAM) and the resulting masks are tracked through subsequent frames with DeAOT, a method of hierarchical propagation for semi-supervised video object segmentation. The geometry of the masks' contours is simplified with the Douglas–Peucker algorithm. Finally, facial traits, pixelation and a basic rim light effect can be optionally added. The results show that the method exhibits an excellent temporal consistency and can correctly process videos with different poses and appearances, dynamic shots, partial shots and diverse backgrounds. The proposed method provides a more simple and deterministic approach than diffusion models based video-to-video translation pipelines, which suffer from temporal consistency problems and do not cope well with pixelated and schematic outputs. The method is also more feasible than techniques based on 3D human pose estimation, which require custom handcrafted 3D models and are very limited with respect to the type of scenes they can process. [ABSTRACT FROM AUTHOR]

Simple Summary: This study presents an automated intelligent technique for real-time identification and assessment of pork belly layers in B-ultrasound images. This non-invasive method can boost the efficiency of breeders in evaluating the layer count within pork belly. By integrating the imaging features of B-ultrasound with a deep learning architecture tailored for image classification, this approach delivers high-precision recognition and categorization of pork belly strata. The findings indicated that the deep learning model adeptly delineated the boundaries between adipose and lean tissues, precisely discerning various layer counts. The system was successfully implemented in a local setting and is now primed for practical deployment. Pork belly, prized for its unique flavor and texture, is often overlooked in breeding programs that prioritize lean meat production. The quality of pork belly is determined by the number and distribution of muscle and fat layers. This study aimed to assess the number of pork belly layers using deep learning techniques. Initially, semantic segmentation was considered, but the intersection over union (IoU) scores for the segmented parts were below 70%, which is insufficient for practical application. Consequently, the focus shifted to image classification methods. Based on the number of fat and muscle layers, a dataset was categorized into three groups: three layers (n = 1811), five layers (n = 1294), and seven layers (n = 879). Drawing upon established model architectures, the initial model was refined for the task of learning and predicting layer traits from B-ultrasound images of pork belly. After a thorough evaluation of various performance metrics, the ResNet18 model emerged as the most effective, achieving a remarkable training set accuracy of 99.99% and a validation set accuracy of 96.22%, with corresponding loss values of 0.1478 and 0.1976. The robustness of the model was confirmed through three interpretable analysis methods, including grad-CAM, ensuring its reliability. Furthermore, the model was successfully deployed in a local setting to process B-ultrasound video frames in real time, consistently identifying the pork belly layer count with a confidence level exceeding 70%. By employing a scoring system with 100 points as the threshold, the number of pork belly layers in vivo was categorized into superior and inferior grades. This innovative system offers immediate decision-making support for breeding determinations and presents a highly efficient and precise method for assessment of pork belly layers. [ABSTRACT FROM AUTHOR]

Today, video analytics are becoming extremely popular due to the increasing need for extracting valuable information from videos available in public sharing services through camera-driven streams in IoT environments. To avoid data communication overheads, a common practice is to have computation close to the data source rather than Cloud offloading. Typically, video analytics are organized as separate tasks, each with different resource requirements (e.g., computational- vs. memory-intensive tasks). The serverless computing paradigm forms a promising approach for mapping such types of applications, enabling fine-grained deployment and management in a per-function, and per-device manner. However, there is a tradeoff between QoS adherence and resource efficiency. Performance variability due to function co-location and prevalent resource heterogeneity make maintaining QoS challenging. At the same time, resource efficiency is essential to avoid waste, such as unnecessary power consumption and CPU reservation. In this paper, we present Darly, a QoS-, interference- and heterogeneity-aware Deep Reinforcement Learning-based Scheduler for serverless video analytics deployments on top of distributed Edge nodes. The proposed framework incorporates a DRL agent that exploits performance counters to identify the levels of interference and the degree of heterogeneity in the underlying Edge infrastructure. It combines this information along with user-defined QoS requirements to improve resource allocations by deciding the placement, migration, or horizontal scaling of serverless functions. We evaluate Darly on a typical Edge cluster with a real-world workflow composed of commonly used serverless video analytics functions and show that our approach achieves efficient scheduling of the deployed functions by satisfying multiple QoS requirements for up to 91.6% (Profile-based) of the total requests under dynamic conditions. [ABSTRACT FROM AUTHOR]

If a ball is incident obliquely on a horizontal surface and is allowed to bounce more than once, then it is likely to bounce many times before it starts rolling along the surface. The number of bounces before rolling commences depends on the initial vertical speed and the normal coefficient of restitution. The transition from bouncing to rolling is examined using a simple theoretical model and is compared with experimental data obtained by filming the process with a video camera. We find that the final rolling speed is proportional to the initial horizontal speed of the ball and depends on the initial ball spin, but is independent of the tangential coefficient of restitution. Representative videos for different balls are included as supplementary material, including a superball thrown with a backspin that creates a back and forth motion. Instructors could use the experiment and/or analysis for an advanced undergraduate lab or use a simplified observational exercise for non-majors. Editor's Note: When a ball is dropped on a horizontal surface with no initial spin, previous studies have found that its bouncing behavior can be simply described using a coefficient of restitution, which gives the ratio of the velocity after the bounce to before the bounce. The value of this coefficient is −1 for a perfectly elastic ball/surface, with a smaller magnitude for any real ball. In many sports, like golf, basketball, or bowling, balls are thrown at an angle and are often given some initial spin by the player. Depending on initial conditions, these balls can bounce, roll, or start by bouncing and then transition to rolling. Here, the transition from bouncing to rolling is shown to be described by using both a vertical and horizontal coefficient of restitution, with the horizontal velocity defined at the spinning edge of the ball rather than its center. Videos of undergraduate level experiments are included, with results used to validate the model. [ABSTRACT FROM AUTHOR]

Violence can be committed anywhere, even in crowded places. It is hence necessary to monitor human activities for public safety. Surveillance cameras can monitor surrounding activities but require human assistance to continuously monitor every incident. Automatic violence detection is needed for early warning and fast response. However, such automation is still challenging because of low video resolution and blind spots. This paper uses ResNet50v2 and the gated recurrent unit (GRU) algorithm to detect violence in the Movies, Hockey, and Crowd video datasets. Spatial features were extracted from each frame sequence of the video using a pretrained model from ResNet50V2, which was then classified using the optimal trained model on the GRU architecture. The experimental results were then compared with wavelet feature extraction methods and classification models, such as the convolutional neural network and long short‐term memory. The results show that the proposed combination of ResNet50V2 and GRU is robust and delivers the best performance in terms of accuracy, recall, precision, and F1‐score. The use of ResNet50V2 for feature extraction can improve model performance. [ABSTRACT FROM AUTHOR]

Darwin proposed that blushing—the reddening of the face owing to heightened self-awareness—is 'the most human of all expressions'. Yet, relatively little is known about the underlying mechanisms of blushing. Theories diverge on whether it is a rapid, spontaneous emotional response that does not involve reflection upon the self or whether it results from higher-order socio-cognitive processes. Investigating the neural substrates of blushing can shed light on the mental processes underlying blushing and the mechanisms involved in self-awareness. To reveal neural activity associated with blushing, 16–20 year-old participants (n = 40) watched pre-recorded videos of themselves (versus other people as a control condition) singing karaoke in a magnetic resonance imaging scanner. We measured participants' cheek temperature increase—an indicator of blushing—and their brain activity. The results showed that blushing is higher when watching oneself versus others sing. Those who blushed more while watching themselves sing had, on average, higher activation in the cerebellum (lobule V) and the left paracentral lobe and exhibited more time-locked processing of the videos in early visual cortices. These findings show that blushing is associated with the activation of brain areas involved in emotional arousal, suggesting that it may occur independently of higher-order socio-cognitive processes. Our results provide new avenues for future research on self-awareness in infants and non-human animals. [ABSTRACT FROM AUTHOR]

Video classification has achieved remarkable success in recent years, driven by advanced deep learning models that automatically categorize video content. This paper provides a comprehensive review of video classification techniques and the datasets used in this field. We summarize key findings from recent research, focusing on network architectures, model evaluation metrics, and parallel processing methods that enhance training speed. Our review includes an in-depth analysis of state-of-the-art deep learning models and hybrid architectures, comparing models to traditional approaches and highlighting their advantages and limitations. Critical challenges such as handling large-scale datasets, improving model robustness, and addressing computational constraints are explored. By evaluating performance metrics, we identify areas where current models excel and where improvements are needed. Additionally, we discuss data augmentation techniques designed to enhance dataset accuracy and address specific challenges in video classification tasks. This survey also examines the evolution of convolutional neural networks (CNNs) in image processing and their adaptation to video classification tasks. We propose future research directions and provide a detailed comparison of existing approaches using the UCF-101 dataset, highlighting progress and ongoing challenges in achieving robust video classification. [ABSTRACT FROM AUTHOR]

Video watermarking is a technique used to embed a visible or invisible mark onto a video file in order to claim ownership, protect intellectual property rights, and detect unauthorized users. The issue of video piracy arises due to the advancement of Internet services and the utilization of various storage technologies. This problem becomes particularly significant with the widespread propagation of media through online channels. The constant progress in sharing of data, videos, images, and applications necessitates the implementation of measures focused on security and the protection of intellectual ownership rights. To address this issue, we proposed a novel Bitstream video watermarking technique that integrates cutting‐edge methodologies to seamlessly embed image data within digital videos. This technique stands apart from less sophisticated watermarking methods by capitalizing on spatial and temporal domain concepts in video processing. An experiment is conducted to analyze the time required for the insertion and extraction processes to gain insights into the performance of bitstream watermarking in the video. While enhancing the existing approach, we systematically explored diverse strategies aimed at augmenting watermark capacity, bolstering robustness, and heightening visual imperceptibility. These strategies encompassed the intricate infusion of watermark fragments into the frame area through chaotic mechanisms, the integration of error‐correcting codes to ensure data fidelity, and the selective omission of dynamic video segments during information extraction processes. Rigorous experimentations are undertaken by the constraints of lossy compression scenarios, maintaining equivalent or reduced bitrate parameters. The findings substantiate the favorable outcomes stemming from the implemented enhancement, affirming the tangible suitability of the phase watermarking technique for the protection of video copyrights. In order to assess the quality of the bitstream video watermarking technique, the peak signal‐to‐noise ratio (PSNR), structural similarity index (SSIM), and normalized correlation coefficient (NCC) are computed and examined by the testing data. The suggested model achieved a PSNR value up to 32.52 dB, SSIM value up to 0.97, and NCC value is 1. The dynamic range (DR) and spatial resolution (SR) are measured in the evaluation process. Security is enhanced significantly after implementing the multifaceted approach. The watermarking scheme is more robust and can effectively resist the potential attacks. In summary, this study advances the field of video watermarking by proposing a sophisticated bitstream technique and rigorously evaluating its performance. The outcomes underscore its practical viability for copyright protection, contributing to a more secure digital media landscape. [ABSTRACT FROM AUTHOR]

Edge computing is essential to handle increasing data volumes and processing capacities. It provides real-time and secure data processing near data sources, like smart devices, alleviating cloud computing energy use, and saving network bandwidth. Specialized accelerators, like GPUs and FPGAs, are vital for low-latency edge computing but the requirements to customized code for different hardware and vendors suppose important compatibility issues. This paper evaluates the potential of SYCL in addressing code portability issues encountered in edge computing. We employed the Polybench suite to compare various SYCL implementations, specifically DPC++ and AdaptiveCpp, with the native solution, CUDA. The disparity between SYCL implementations was negligible, at just 5%. Furthermore, we evaluated SYCL in the context of specific edge computing applications such as video processing using three different optical flow algorithms. The results revealed a slight performance gap of 3% when transitioning from CUDA to SYCL. Upon evaluating energy consumption, the observed difference ranged from ± 10 % , depending on the application utilized. These gaps are the price one may need to pay when achieving the ability to successfully run the same code on two distinct edge boards. These findings underscore SYCL's capacity to increase productivity in terms of development costs and facilitate IoT deployment without being locked into a particular platform or manufacturer. [ABSTRACT FROM AUTHOR]

The digitization of production systems has revolutionized industrial monitoring. Analyzing real-time bottom-up data enables the dynamic monitoring of industrial processes. Data are collected in various types, like video frames and time signals. This article focuses on leveraging images from a vision system to monitor the manufacturing process on a computer numerical control (CNC) lathe machine. We propose a method for designing and integrating these video modules on the edge of a production line. This approach detects the presence of raw parts, measures process parameters, assesses tool status, and checks roughness in real time using image processing techniques. The efficiency is evaluated by checking the deployment, the accuracy, the responsiveness, and the limitations. Finally, a perspective is offered to use the metadata off the edge in a more complex artificial-intelligence (AI) method for predictive maintenance. [ABSTRACT FROM AUTHOR]

This paper introduces a new descriptor called orientation‐magnitude response maps as a single 2D image to effectively explore motion patterns. Moreover, boosted multi‐stream CNN‐based model with various attention modules is designed for human action recognition. The model incorporates a convolutional self‐attention autoencoder to represent compressed and high‐level motion features. Sequential convolutional self‐attention modules are used to exploit the implicit relationships within motion patterns. Furthermore, 2D discrete wavelet transform is employed to decompose RGB frames into discriminative coefficients, providing supplementary spatial information related to the actors actions. A spatial attention block, implemented through the weighted inception module in a CNN‐based structure, is designed to weigh the multi‐scale neighbours of various image patches. Moreover, local and global body pose features are combined by extracting informative joints based on geometry features and joint trajectories in 3D space. To provide the importance of specific channels in pose descriptors, a multi‐scale channel attention module is proposed. For each data modality, a boosted CNN‐based model is designed, and the action predictions from different streams are seamlessly integrated. The effectiveness of the proposed model is evaluated across multiple datasets, including HMDB51, UTD‐MHAD, and MSR‐daily activity, showcasing its potential in the field of action recognition. [ABSTRACT FROM AUTHOR]

Video captioning aims to identify multiple objects and their behaviours in a video event and generate captions for the current scene. This task aims to generate a detailed description of the current video in real‐time using natural language, which requires deep learning to analyze and determine the relationships between interesting objects in the frame sequence. In practice, existing methods typically involve detecting objects in the frame sequence and then generating captions based on features extracted through object coverage locations. Therefore, the results of caption generation are highly dependent on the performance of object detection and identification. This work proposes an advanced video captioning approach that works in adaptively and effectively addresses the interdependence between event proposals and captions. Additionally, an attention‐based multimodel framework is introduced to capture the main context from the frame and sound in the video scene. Also, an intermediate model is presented to collect the hidden states captured from the input sequence, which performs to extract the main features and implicitly produce multiple event proposals. For caption prediction, the proposed method employs the CARU layer with attention consideration as the primary RNN layer for decoding. Experimental results showed that the proposed work achieves improvements compared to the baseline method and also better performance compared to other state‐of‐the‐art models on the ActivityNet dataset, presenting competitive results in the tasks of video captioning. [ABSTRACT FROM AUTHOR]

Public authorities and private companies have used video cameras as part of surveillance systems, and one of their objectives is the rapid detection of physically violent actions. This task is usually performed by human visual inspection, which is labor-intensive. For this reason, different deep learning models have been implemented to remove the human eye from this task, yielding positive results. One of the main problems in detecting physical violence in videos is the variety of scenarios that can exist, which leads to different models being trained on datasets, leading them to detect physical violence in only one or a few types of videos. In this work, we present an approach for physical violence detection on images obtained from video based on threshold active learning, that increases the classifier's robustness in environments where it was not trained. The proposed approach consists of two stages: In the first stage, pre-trained neural network models are trained on initial datasets, and we use a threshold (μ) to identify those images that the classifier considers ambiguous or hard to classify. Then, they are included in the training dataset, and the model is retrained to improve its classification performance. In the second stage, we test the model with video images from other environments, and we again employ (μ) to detect ambiguous images that a human expert analyzes to determine the real class or delete the ambiguity on them. After that, the ambiguous images are added to the original training set and the classifier is retrained; this process is repeated while ambiguous images exist. The model is a hybrid neural network that uses transfer learning and a threshold μ to detect physical violence on images obtained from video files successfully. In this active learning process, the classifier can detect physical violence in different environments, where the main contribution is the method used to obtain a threshold μ (which is based on the neural network output) that allows human experts to contribute to the classification process to obtain more robust neural networks and high-quality datasets. The experimental results show the proposed approach's effectiveness in detecting physical violence, where it is trained using an initial dataset, and new images are added to improve its robustness in diverse environments. [ABSTRACT FROM AUTHOR]

The challenge of multi-object tracking stands as a fundamental focus in computer vision research, finding widespread applications in areas such as public safety, transportation, autonomous vehicles, robotics, and other domains involving artificial intelligence. Given the intricate nature of natural scenes, the occurrence of object occlusion and semi-occlusion is commonplace in basic tracking tasks. These factors often result in challenges such as ID switching, object loss, detection errors, and misaligned bounding boxes, thereby significantly impacting the precision of multi-object tracking.This paper aims to address the aforementioned issues and proposes a novel multi-object tracker, incorporating Relative location mapping (RLM) and Target region density (TRD) modeling. The new tracker is more sensitive to differences in the spatial relationships between targets, allowing it to dynamically introduce low-scoring detection boxes into different regions based on the density of target regions in the image. This improves the accuracy of target tracking while avoiding the consumption of a significant amount of computational resources.Our research results indicate that when applying this method to state-of-the-art multi-object tracking approaches, the proposed model achieves improvements of 0.4 to 0.8 points in the HOTA and IDF1 metrics on the MOT17 and MOT20 datasets. This demonstrates the effectiveness of the proposed method in enhancing multi-object tracking performance. [ABSTRACT FROM AUTHOR]

With the continuous development of Internet technology, using multimedia for virtual application has become a new way. In this paper, by introducing the bayesian classification algorithm, multimedia graphics and video for carding, testing and implementation of edge image of fine processing, matching the corresponding text for quick positioning, at the same time use the filter for image and video background picture, further classification, distinguish between background and text, the extraction of superposition character. Simulation results show that the bayesian classification algorithm is effective and improves the efficiency and accuracy of image and video processing. [ABSTRACT FROM AUTHOR]

In-loop restoration is a post-processing task aiming to reduce losses caused by the quantization and the inverse quantization phases in a video coding process. Emerging in-loop restoration methods, most of them based on deep learning, have reported higher quality gains than classical filters. However, the complexity at the decoder side remains a challenge. The Sparse Restoration Method (SRM) is presented as a low-complexity method that utilizes sparse representation and Natural Scene Statistic metrics to enhance visual quality at the block level. Our method shows potential restoration benefits when applied to synthetic video sequences. [ABSTRACT FROM AUTHOR]

This paper introduces a variational autoencoder network designed for video colorization using reference images, addressing the challenge of colorizing black-and-white videos. Although recent techniques perform well in some scenarios, they often struggle with color inconsistencies and artifacts in videos that feature complex scenes and long durations. To tackle this, we propose a variational autoencoder framework that incorporates spatio-temporal information for efficient video colorization. To improve temporal consistency, we unify semantic correspondence with color propagation, allowing for simultaneous guidance in colorizing grayscale video frames. Additionally, the variational autoencoder learns spatio-temporal feature representations by mapping video frames into a latent space through an encoder network. The decoder network then transforms these latent features back into color images. Compared to traditional coloring methods, our approach accurately captures temporal relationships between video frames, providing precise colorization while ensuring video consistency. To further enhance video quality, we apply a specialized loss function that constrains the generated output, ensuring that the colorized video remains spatio-temporally consistent and natural. Experimental results demonstrate that our method significantly improves the video colorization process. [ABSTRACT FROM AUTHOR]

High‐angle cameras are commonly used for trajectory data collection in transportation research. However, without refinement and validation, trajectory data obtained through video processing software may be unreliable, inaccurate, or incomplete. This paper focuses on a critical issue in the field of trajectory data acquisition and analysis—there is still no reliable and fully vetted trajectory dataset in the research community. The current practice for validating video‐based trajectory can be classified as indirect methods and direct methods. Indirect methods of trajectory validation use algorithms to efficiently correct data anomalies without human intervention but may overlook detailed driving behaviors, whereas direct methods involve meticulous manual verification to preserve data fidelity but are labor‐intensive and less scalable. The spatial‐temporal maps (STMaps) method offers an additional layer of verification to affirm the accuracy and reliability of trajectory data. To enhance the performance, the deep spatial‐temporal embedding model is proposed for trajectory instance segmentation on STMaps using the contrastive learning framework. The parity constraints at both pixel and instance levels guide the deep neural network to learn the embedding spaces that can be built on different backbone networks. The reconstructed Next Generation Simulation (NGSIM) highway dataset trajectory dataset is thoroughly validated against manually processed ground truth, and the error‐free NGSIM data are refined to be a reliable resource for transportation research based on car‐following behaviors, lane‐change frequency, consistency, and jerk value measurements. [ABSTRACT FROM AUTHOR]

Most current deep learning models are suboptimal in terms of the flexibility of their input shape. Usually, computer vision models only work on one fixed shape used during training, otherwise their performance degrades significantly. For video-related tasks, the length of each video (i.e., number of video frames) can vary widely; therefore, sampling of video frames is employed to ensure that every video has the same temporal length. This training method brings about drawbacks in both the training and testing phases. For instance, a universal temporal length can damage the features in longer videos, preventing the model from flexibly adapting to variable lengths for the purposes of on-demand inference. To address this, we propose a simple yet effective training paradigm for 3D convolutional neural networks (3D-CNN) which enables them to process videos with inputs having variable temporal length, i.e., variable length training (VLT). Compared with the standard video training paradigm, our method introduces three extra operations during training: sampling twice, temporal packing, and subvideo-independent 3D convolution. These operations are efficient and can be integrated into any 3D-CNN. In addition, we introduce a consistency loss to regularize the representation space. After training, the model can successfully process video with varying temporal length without any modification in the inference phase. Our experiments on various popular action recognition datasets demonstrate the superior performance of the proposed method compared to conventional training paradigm and other state-of-the-art training paradigms. [ABSTRACT FROM AUTHOR]

The real-time tracking of moving objects has extensive applications in various domains. Existing tracking methods typically utilize video image processing, but their performance is limited due to the high information throughput and computational requirements associated with processing continuous images. Additionally, imaging in certain spectral bands can be costly. This paper proposes a non-imaging real-time three-dimensional tracking technique for distant moving targets using single-pixel LiDAR. This novel approach involves compressing scene information from three-dimensional to one-dimensional space using spatial encoding modulation and then obtaining this information through single-pixel detection. A LiDAR system is constructed based on this method, where the peak position of the detected full-path one-dimensional echo signal is used to obtain the target distance, while the peak intensity is used to obtain the azimuth and pitch information of the moving target. The entire process requires minimal data collection and a low computational load, making it feasible for the real-time three-dimensional tracking of single or multiple moving targets. Outdoor experiments confirmed the efficacy of the proposed technology, achieving a distance accuracy of 0.45 m and an azimuth and pitch angle accuracy of approximately 0.03° in localizing and tracking a flying target at a distance of 3 km. [ABSTRACT FROM AUTHOR]

Approximate computing has emerged as a design alternative to enhance design efficiency by capitalizing on the inherent error resilience observed in numerous applications. Various error-resilient and compute-intensive applications, such as signal, image and video processing, computer vision, and supervised machine learning, necessitate dedicated hardware accelerators for mean squared error estimation during runtime. In these application domains, using efficient arithmetic operators, particularly a squarer unit, represents one of the most effective strategies for low-power design. This work introduces an approximate Radix- 2 m squarer unit, denoted as AxRSU- 2 m . The proposed squarer unit employs m-bit approximate encoders to execute operations on m-bit data concurrently. The AxRSU- 2 m under consideration explores encoders with m equal to 2 (AxRSU-4), 3 (AxRSU-8), and 4 (AxRSU-16). These approximate encoders exhibit low complexity and diminish the necessary partial products operating on m bits simultaneously, thereby substantially enhancing energy efficiency and reducing circuit area in the AxRSU- 2 m . To illustrate the trade-off between error and quality in the AxRSU- 2 m , we apply it to an SSD (sum squared difference) hardware accelerator designed for video processing, with a square-accumulate serving as a case study. Our findings reveal a novel Pareto front, presenting eight optimal AxRSU- 2 m solutions that achieve accuracy levels ranging from 3.76 to 75.53%. These solutions yield energy savings ranging from 46.20 to 95.57% and circuit area reductions ranging from 37.68 to 66.73%. [ABSTRACT FROM AUTHOR]

Video frame interpolation (VFI) is a technique that synthesises intermediate frames between adjacent original video frames to enhance the temporal super‐resolution of the video. However, existing methods usually rely on heavy model architectures with a large number of parameters. The authors introduce an efficient VFI network based on multiple lightweight convolutional units and a Local three‐scale encoding (LTSE) structure. In particular, the authors introduce a LTSE structure with two‐level attention cascades. This design is tailored to enhance the efficient capture of details and contextual information across diverse scales in images. Secondly, the authors introduce recurrent convolutional layers (RCL) and residual operations, designing the recurrent residual convolutional unit to optimise the LTSE structure. Additionally, a lightweight convolutional unit named separable recurrent residual convolutional unit is introduced to reduce the model parameters. Finally, the authors obtain the three‐scale decoding features from the decoder and warp them for a set of three‐scale pre‐warped maps. The authors fuse them into the synthesis network to generate high‐quality interpolated frames. The experimental results indicate that the proposed approach achieves superior performance with fewer model parameters. [ABSTRACT FROM AUTHOR]

A conventional codec aims to increase the compression efficiency for transmission and storage while maintaining video quality. However, as the number of platforms using machine vision rapidly increases, a codec that increases the compression efficiency and maintains the accuracy of machine vision tasks must be devised. Hence, the Moving Picture Experts Group created a standardization process for video coding for machines (VCM) to reduce bitrates while maintaining the accuracy of machine vision tasks. In particular, in‐loop filters have been developed for improving the subjective quality and machine vision task accuracy. However, the high computational complexity of in‐loop filters limits the development of a high‐performance VCM architecture. We analyze the effect of an in‐loop filter on the VCM performance and propose a suboptimal VCM method based on the selective activation of in‐loop filters. The proposed method reduces the computation time for video coding by approximately 5% when using the enhanced compression model and 2% when employing a Versatile Video Coding test model while maintaining the machine vision accuracy and compression efficiency of the VCM architecture. [ABSTRACT FROM AUTHOR]

The paper introduces the Integrated Measurement of Performance and Contextual Task-related effects (IMPACT) dataset, designed to facilitate comprehensive investigations into contextual dynamics within human–machine collaboration (HMC) systems. Unlike traditional approaches focusing solely on monitoring physical contextual aspects, IMPACT uniquely incorporates the individual perception of human participants within these systems. Data sources include temporal records of task performance, user models derived from real-time video processing as well as contextual factors such as different light and noise conditions. Key features include task stimuli, user responses, reaction times, emotional states, and attention levels, all synchronized via timestamps and recorded in.csv and.mp4 formats. Our analysis highlights variations in user perceptions and performance under different contextual states, both in person-independent and person-specific scenarios. Pre-test and post-test questionnaire data reveal shifts in user perceptions of light and noise as distractors. Performance data indicate that task-related adaptations maintain consistent performance levels despite contextual changes, while attention and arousal levels vary significantly. Person-specific analysis underscores the importance of individualized context adaptation, as users exhibit unique responses to environmental changes. The IMPACT dataset supports the development of adaptive human–machine collaboration systems by integrating individual user perceptions with objective context monitoring. Future research will focus on refining context-adaptive models to enhance the robustness and accuracy of individualized context-related performance predictions. [ABSTRACT FROM AUTHOR]

Aiming to automatically monitor and improve stereoscopic image and video processing systems, stereoscopic image quality assessment approaches are becoming more and more important as 3D technology gains popularity. We propose a fullreference stereoscopic image quality assessment method that incorporate monocular and binocular features based on binocular competition and binocular integration. To start, we create a three-channel RGB fused view by fusing Gabor filter bank responses and disparity maps. Then, using the monocular view and the RGB fusion view, respectively, we extract monocular and binocular features. To alter the local features in the binocular features, we simultaneously estimate the saliency of the RGB fusion image. Finally, the monocular and binocular quality scores are calculated based on the monocular and binocular features, and the quality scores of the stereo image prediction are obtained by fusion. Performance testing in the LIVE 3D IQA database Phase I and Phase II. The results of the proposed method are compared with newer methods. The experimental results show good consistency and robustness. [ABSTRACT FROM AUTHOR]

Searching for objects is a common task in daily life and work. For augmented reality (AR) devices without spatial perception systems, the image of the object's last appearance serves as a common search assistance. Compared to using only images as visual cues, videos capturing the process of object placement can provide procedural guidance, potentially enhancing users' search efficiency. However, complete video playback capturing the entire object placement process as visual cues can be excessively lengthy, requiring users to invest significant viewing time. To explore whether segmented or accelerated video playback can still assist users in object retrieval tasks effectively, we conducted a user study. The results indicated that when video playback is covering the first appearance of the object's destination to the object's final appearance (referred to as the destination appearance, DA) and playing at normal speed, search time and cognitive load were significantly reduced. Subsequently, we designed a second user study to evaluate the performance of video playback compared to image cues in object retrieval tasks. The results showed that combining the DA playback starting point with images of the object's last appearance further reduced search time and cognitive load. [ABSTRACT FROM AUTHOR]

Online detection of action start is a significant and challenging task that requires prompt identification of action start positions and corresponding categories within streaming videos. This task presents challenges due to data imbalance, similarity in boundary content, and real‐time detection requirements. Here, a novel Time‐Attentive Fusion Network is introduced to address the requirements of improved action detection accuracy and operational efficiency. The time‐attentive fusion module is proposed, which consists of long‐term memory attention and the fusion feature learning mechanism, to improve spatial‐temporal feature learning. The temporal memory attention mechanism captures more effective temporal dependencies by employing weighted linear attention. The fusion feature learning mechanism facilitates the incorporation of current moment action information with historical data, thus enhancing the representation. The proposed method exhibits linear complexity and parallelism, enabling rapid training and inference speed. This method is evaluated on two challenging datasets: THUMOS'14 and ActivityNet v1.3. The experimental results demonstrate that the proposed method significantly outperforms existing state‐of‐the‐art methods in terms of both detection accuracy and inference speed. [ABSTRACT FROM AUTHOR]