Search

Recognizing actions from still images has been popularly studied recently. In this paper, we model an action class as a flexible number of spatial configurations of body parts by proposing a new spatial sum–product network (SPN). First, we discover a set of parts in image collections via unsupervised learning. Then, our new spatial SPN is applied to model the spatial relationship and also the high-order correlations of parts. To learn robust networks, we further develop a hierarchical spatial SPN method, which models pairwise spatial relationship between parts inside subimages and models the correlation of subimages via extra layers of SPN. Our method is shown to be effective on two benchmark data sets. [ABSTRACT FROM PUBLISHER]

Prospective authors are requested to submit new, unpublished manuscripts for inclusion in the upcoming event described in this call for papers. [ABSTRACT FROM AUTHOR]

The article offers information on call for papers IEEE Trans. on circuits and systems for video technology special section on video and Language.

This paper presents a new objective quality metric that was adapted to the complex characteristics of immersive video (IV) which is prone to errors caused by processing and compression of multiple input views and virtual view synthesis. The proposed metric, IV-PSNR, contains two techniques that allow for the evaluation of quality loss for typical immersive video distortions: corresponding pixel shift and global component difference. The performed experiments compared the proposal with 31 state-of-the-art quality metrics, showing their performance in the assessment of quality in immersive video coding and processing, and in other applications, using commonly used image quality assessment databases– TID2013 and CVIQ. As presented, IV-PSNR outperforms other metrics in immersive video applications and still can be efficiently used in the evaluation of different images and videos. Moreover, basing the metric on the calculation of PSNR allowed the computational complexity to remain low. Publicly available, efficient implementation of IV-PSNR software was provided by the authors of this paper and is used by ISO/IEC MPEG for evaluation and research on the upcoming MPEG Immersive video (MIV) coding standard. [ABSTRACT FROM AUTHOR]

This paper presents the overview and rationale behind the Decoder-Side Depth Estimation (DSDE) mode of the MPEG Immersive Video (MIV) standard, using the Geometry Absent profile, for efficient compression of immersive multiview video. A MIV bitstream generated by an encoder operating in the DSDE mode does not include depth maps. It only contains the information required to reconstruct them in the client or in the cloud: decoded views and metadata. The paper explains the technical details and techniques supported by this novel MIV DSDE mode. The description additionally includes the specification on Geometry Assistance Supplemental Enhancement Information which helps to reduce the complexity of depth estimation, when performed in the cloud or at the decoder side. The depth estimation in MIV is a non-normative part of the decoding process, therefore, any method can be used to compute the depth maps. This paper lists a set of requirements for depth estimation, induced by the specific characteristics of the DSDE. The depth estimation reference software, continuously and collaboratively developed with MIV to meet these requirements, is presented in this paper. Several original experimental results are presented. The efficiency of the DSDE is compared to two MIV profiles. The combined non-transmission of depth maps and efficient coding of textures enabled by the DSDE leads to efficient compression and rendering quality improvement compared to the usual encoder-side depth estimation. Moreover, results of the first evaluation of state-of-the-art multiview depth estimators in the DSDE context, including machine learning techniques, are presented. [ABSTRACT FROM AUTHOR]

Recently, research on six-degree-of-freedom virtual reality (VR) systems based on a light field (LF) has been actively conducted. The LF-based approach is photorealistic and less prone to errors than the existing three-dimensional (3D) modeling-based approach. On the other hand, because the amount of data is very large for LF-based immersive virtual reality, it is important to manage the data efficiently. This paper makes two major contributions to data management for rendering acceleration in this context. First, the GPUs commonly used for high-speed rendering require optimization of the data transfer process between the host memory and the device memory. This paper proposes a simple but effective way to maximize data reuse by finding the proper size of a transfer unit depending on the target applications and the system capability constraints. Also, view synthesis and data transfer are performed in a pipelined manner to hide the data movement time. Second, for the LF rendering of a wide space, data management in the storage-host-GPU hierarchy is attempted for the first time. As the size of the virtual space increases, the required amount of LF data grows rapidly and exceeds the allocatable host memory capacity. Because the loading of data stored in storage is a very slow process, a system-level LF rendering structure that carefully considers the characteristics of the storage-host-GPU memory hierarchy should be designed. In this paper, progressive LF updating is proposed, and it prevents rendering stalls caused by slow storage data loading. Experimental results show that a 360-degree view with a $9000 \times 2048$ resolution can be rendered in an average of 2.57 ms from LF data with a $4096 \times 2048$ resolution, achieving performance close to 400 frames per second. [ABSTRACT FROM AUTHOR]

In this paper, we focus on subjective and objective Point Cloud Quality Assessment (PCQA) in an immersive environment and study the effect of geometry and texture attributes in compression distortion. Using a Head-Mounted Display (HMD) with six degrees of freedom, we establish a subjective PCQA database, named SIAT Point Cloud Quality Database (SIAT-PCQD). Our database consists of 340 distorted point clouds compressed by the MPEG point cloud encoder with the combination of 20 sequences and 17 pairs of geometry and texture quantization parameters. The impact of distorted geometry and texture attributes is further discussed in this paper. Then, we propose two projection-based objective quality evaluation methods, i.e., a weighted view projection based model and a patch projection based model. Our subjective database and findings can be used in point cloud processing, transmission, and coding, especially for virtual reality applications. The subjective dataset https://dx.doi.org/10.21227/ad8d-7r28 http://codec.siat.ac.cn/video%5fdownload%5fsiat-pcqd.htmlhas been released in the public repository. [ABSTRACT FROM AUTHOR]

The existing state-of-the-art (SOTA) video salient object detection (VSOD) models have widely followed short-term methodology, which dynamically determines the balance between spatial and temporal saliency fusion by solely considering the current consecutive limited frames. However, the short-term methodology has one critical limitation, which conflicts with the real mechanism of our visual system — a typical long-term methodology. As a result, failure cases keep showing up in the results of the current SOTA models, and the short-term methodology becomes the major technical bottleneck. To solve this problem, this paper proposes a novel VSOD approach, which performs VSOD in a complete long-term way. Our approach converts the sequential VSOD, a sequential task, to a data mining problem, i.e., decomposing the input video sequence to object proposals in advance and then mining salient object proposals as much as possible in an easy-to-hard way. Since all object proposals are simultaneously available, the proposed approach is a complete long-term approach, which can alleviate some difficulties rooted in conventional short-term approaches. In addition, we devised an online updating scheme that can grasp the most representative and trustworthy pattern profile of the salient objects, outputting framewise saliency maps with rich details and smoothing both spatially and temporally. The proposed approach outperforms almost all SOTA models on five widely used benchmark datasets. [ABSTRACT FROM AUTHOR]

This paper presents a novel method to determine rate-distortion optimized transform coefficients for efficient compression of videos generated from point clouds. The method exploits a generalized frequency selective extrapolation approach that iteratively determines rate-distortion-optimized coefficients for all basis functions of two-dimensional discrete cosine and sine transforms. The method is applied to blocks containing both occupied and unoccupied pixels in video based point cloud compression for HEVC encoding. In the proposed algorithm, only the values of the transform coefficients are changed such that resulting bit streams are compliant to the V-PCC standard. For all-intra coded point clouds, bitrate savings of more than 4% for geometry and more than 6% for texture error metrics with respect to standard encoding can be observed. These savings are more than twice as high as savings obtained using competing methods from literature. In the randomaccess case, our proposed method outperforms competing V-PCC methods by more than 0.5%. [ABSTRACT FROM AUTHOR]

Video summarization (VS) is generally formulated as a subset selection problem where a set of representative keyframes or key segments is selected from an entire video frame set. Though many sparse subset selection based VS algorithms have been proposed in the past decade, most of them adopt linear sparse formulation in the explicit feature vector space of video frames, and don’t consider the local or global relationships among frames. In this paper, we first extend the conventional sparse subset selection for VS into kernel block sparse subset selection (KBS3) to utilize the advantage of kernel sparse coding and introduce a local inter-frame relationship through packing of frame blocks. Going a step further, we propose a similarity based block sparse subset selection (SB2S3) model by applying a specially designed transformation matrix on the KBS3 model in order to introduce a kind of global inter-frame relationship through the similarity. Finally, a greedy pursuit based algorithm is devised for the proposed NP-hard model optimization. The proposed SB2S3 has the following advantages: 1) through the similarity between each frame and any other frame, the global relationship among all frames can be considered; 2) through block sparse coding, the local relationship of adjacent frames is further considered; and 3) it has a wider application, since features can derive similarity, but not vice versa. It is believed that the effect of modeling such global and local relationships among frames in this paper, is similar to that of modeling the long-range and short-range dependencies among frames in deep learning based methods. Experimental results on three benchmark datasets have demonstrated that the proposed approach is superior to not only other sparse subset selection based VS methods but also most unsupervised deep-learning based VS methods. [ABSTRACT FROM AUTHOR]

Video grounding aims at localizing the temporal moment related to the given language description, which is very helpful to many cross-modal content understanding applications like visual question answering and sentence-video search. Existing approaches usually directly regress the temporal boundaries of an event described by a query sentence in the video sequence. This direct regression manner often encounters a large decision space due to diverse target events and variable video durations, leading to inaccurate localization as well as inefficient grounding. This paper presents an efficient framework termed from which to where to facilitate video grounding. The core idea is imitating the reading comprehension process to gradually narrow the decision space, in what we decompose the direct regression into two steps. The “which” step first roughly selects a candidate area by evaluating which video segment in the predefined set is closest to the ground truth. To this end, we formulate this step into a multi-choice reading comprehension problem and propose a criterion to select the best-matched segment. In this way, the excessive decision space is effectively reduced. The “where” step aims to precisely regress the temporal boundary of the selected video segment from the shrunk decision space. We thus introduce a triple-span representation for each candidate video segment to use the regional context for better boundary regression. The “which” and “where” steps can be combined into a unified framework and learned end-to-end, leading to an efficient video grounding system. Extensive experiments on Charades-STA, ActivityNet-Captions, and TACoS benchmarks clearly demonstrate the effectiveness of our framework. [ABSTRACT FROM AUTHOR]

Video saliency prediction (VSP) aims to imitate eye fixations of humans. However, the potential of this task has not been fully exploited since existing VSP methods only focus on modeling visual saliency of the input previous frames. In this paper, we present the first attempt to extend this task to video saliency forecasting (VSF) by forecasting attention regions of consecutive future frames. To tackle this problem, we propose a video saliency forecasting transformer (VSFT) network built on a new encoder-decoder architecture. Different from existing VSP methods, our VSFT is the first pure-transformer based architecture in the VSP field and is freed from the dependency of the pretrained S3D model. In VSFT, the attention mechanism is exploited to capture spatial-temporal dependencies between the input past frames and the target future frame. We propose cross-attention guidance blocks (CAGB) to aggregate multi-level representation features to provide sufficient guidance for forecasting. We conduct comprehensive experiments on two benchmark datasets, DHF1K and Hollywoods-2. We investigate the saliency forecasting and predicting abilities of existing VSP methods by modifying the supervision signals. Experimental results demonstrate that our method achieves superior performance on both VSF and VSP tasks. [ABSTRACT FROM AUTHOR]

Attention mechanisms have significantly boosted the performance of video classification neural networks thanks to the utilization of perspective contexts. However, the current research on video attention generally focuses on adopting a specific aspect of contexts (e.g., channel, spatial/temporal, or global context) to refine the features and neglects their underlying correlation when computing attentions. This leads to incomplete context utilization and hence bears the weakness of limited performance improvement. To tackle the problem, this paper proposes an efficient attention-in-attention (AIA) method for element-wise feature refinement, which investigates the feasibility of inserting the channel context into the spatio-temporal attention learning module, referred to as CinST, and also its reverse variant, referred to as STinC. Specifically, we instantiate the video feature contexts as dynamics aggregated along a specific axis with global average and max pooling operations. The workflow of an AIA module is that the first attention block uses one kind of context information to guide the gating weights calculation of the second attention that targets at the other context. Moreover, all the computational operations in attention units act on the pooled dimension, which results in quite few computational cost increase (< 0.02%). To verify our method, we densely integrate it into two classical video network backbones and conduct extensive experiments on several standard video classification benchmarks. The source code of our AIA is available at https://github.com/haoyanbin918/Attention-in-Attention. [ABSTRACT FROM AUTHOR]

Video captioning is a joint task of computer vision and natural language processing, which aims to describe the video content using several natural language sentences. Nowadays, most methods cast this task as a mapping problem, which learns a mapping from visual features to natural language and generates captions directly from videos. However, the underlying challenge of video captioning, i.e., sequence to sequence mapping across the different domains, is still not well handled. To address these problems, we introduce the polishing mechanism in an attempt to mimic human polishing process and propose a generate-and-polish framework for video captioning. In this paper, we propose a two-step transformer based polishing network (TSTPN) consisting of two sub-modules: the generation-module is to generate the caption candidate and the polishing-module is to gradually refine the generated candidate. Specifically, the candidate provides a global information of the visual contents in a semantically-meaningful order, where it is firstly considered as a semantic intersnubber to bridge the semantic gap between the text and video, with the cross-modal attention mechanism for better cross-modal modeling; and it secondly provides a global planning ability to maintain the semantic consistency and fluency of the whole sentence for better sequence mapping. In experiments, we present adequate evaluations to show that the proposed TSTPN achieves the comparable and even better performance than the state-of-the-art methods on the benchmark datasets. [ABSTRACT FROM AUTHOR]

Panoramic video contains 360-degree video content and it is convenient to draw the content that the user wants to watch on the head-Mount display (HMD) of the virtual reality display device. However, few encoders code panoramic video directly. Panoramic video content needs to be projected to a compression-friendly rectangular plane first. Then the existing video coding standards such as AVC, HEVC, VVC, AVS are used to code the panoramic video content. In order to improve the panoramic video coding efficiency, IEEE 1857.9 immersive video content coding working group was established to develop efficient immersive video projection and coding methods. This paper presents an overview of the panoramic video projection schemes in Immersive Visual Content Coding Standard IEEE 1857.9. [ABSTRACT FROM AUTHOR]

We observed that remarkable and impressive performance on image-based human pose estimation have been achieved by deep Convolutional Neural Networks (CNN). Nevertheless, directly applying these image-based models on videos is not only computionally intensive, but also may cause jitter and loss. The main reason is that the image-based models purely focus on the local features of individual frames and totally ignore the temporal information among adjacent frames. Some existing methods are proposed to address the temporal coherency issue. However, these methods need to be designed carefully and cannot be combined with existing image-based methods. In this paper, we propose a simple yet effective module to refine the estimated pose by exploiting the temporal coherency among the heatmaps of adjacent frames, which can be easily inserted into image-based networks as a plug-in. We show that the temporal coherency issue among the heatmap frames could be re-formulated as a graph path selection optimization problem. Moreover, to speed up the refinement process, we propose a hierarchical graph optimization to achieve the refinement from coarse to fine. Experimental results on two large-scale video pose estimation benchmarks show that our module can improve the performance with little speed loss when combined with image-based methods as an efficient plug-in. [ABSTRACT FROM AUTHOR]

The onset and progression of Parkinson’s disease (PD) gradually affect the patient’s motor functions and quality of life. The PD motor symptoms are usually assessed using the Movement Disorder Society-sponsored revision of the Unified Parkinson’s Disease Rating Scale (MDS-UPDRS). Automated MDS-UPDRS assessment has been recently required as an invaluable tool for PD diagnosis and telemedicine, especially with the recent novel coronavirus pandemic outbreak. This paper proposes a novel vision-based method for automated assessment of the arising-from-chair task, which is one of the key MDS-UPDRS components. The proposed method is based on a self-supervised metric learning scheme with a graph convolutional network (SSM-GCN). Specifically, for human skeleton sequences extracted from videos, a self-supervised intra-video quadruplet learning strategy is proposed to construct a metric learning formulation with prior knowledge, for improving the spatial-temporal representations. Afterwards, a vertex-specific convolution operation is designed to achieve effective aggregation of all skeletal joint features, where each joint or feature is weighted differently based on its relative factor of importance. Finally, a graph representation supervised mechanism is developed to maximize the potential consistency between the joint and bone information streams. Experimental results on a clinical dataset demonstrate the superiority of the proposed method over the existing sensor-based methods, with an accuracy of 70.60% and an acceptable accuracy of 98.65%. The analysis of discriminative spatial connections makes our predictions more clinically interpretable. This method can achieve reliable automated PD assessment using only easily-obtainable videos, thus providing an effective tool for real-time PD diagnosis or remote continuous monitoring. [ABSTRACT FROM AUTHOR]

Temporal moment localization using natural language (TMLNL) is an emerging issue in computer vision for localizing a specific moment inside a long, untrimmed video. The goal of TMLNL is to obtain the video’s output moment, which is related to the input query in a substantial way. Previous research focused on the visual portion of TMLNL, such as objects, backdrops, and other visual attributes, but natural language processing (NLP) techniques were largely used for the textual portion. A long query requires sufficient context to properly localize moments within a long untrimmed video. Thus, as a consequence of not completely understanding how to handle queries, performances deteriorated, especially when the query was longer. In this paper, we treat the TMLNL challenge as a unique variation of VQA, which equally considers the visual elements by using our proposed VQA joint visual-textual framework (JVTF). However, we also manage complex and long input queries without employing natural language processing (NLP) by improving poorly graded to finely graded distinct granularity representations. Our suggested BCPN searches for insufficient context for long input queries using an approach called query handler (QH) and helps the JVTF find the most relevant moment. Previously, a recurrence of words was caused by increasing the number of encoding layers in transformers, LSTMs, and other NLP techniques; however, our QH ensured that repetition of word locations was reduced. The output of BCPN is combined with JVTF’s guided attention to further improve the end outcome. Therefore, we propose a novel bidirectional context predictor network (BCPN), in addition to a VQA joint visual-textual framework (JVTF), to address the equal importance of videos and queries. Through extensive experiments on three benchmark datasets, we show that the proposed BCPN outperforms the state-of-the-art methods by $IoU = 0.3 (2.65 \%) $ , $IoU = 0.5 (2.49 \%)$ , and $IoU = 0.7 (2.06 \%) $. [ABSTRACT FROM AUTHOR]

In this paper, we propose an ambiguousness-aware state evolution (AASE) method which represents the uncertainty of the input sequence and evolves the subsequent skeletons to generate a reasonable full-length sequence for action prediction. Unlike most existing methods that enforce partial sequences with the labels of full-length videos and ignore the semantic information of the subsequent action, we develop an evolution method by predicting the instructional actions and generating the reasonable candidate subsequent actions, so that the ambiguity of the full sequence’s label supervising for the partial actions can be effectively alleviated. Our method generates the rational subsequent actions under the instructional action class to complement the partially observed action sequence. We design two criteria for a rational generation: 1) the instruction of subsequent action keeps the semantic consistency with the observed sequence; 2) the generation sequence is satisfied with the distribution of the sequence of real data. Moreover, we design an uncertainty module to decide the instructional action class for the generation network. AASE predicts instructional actions with uncertainty learning and evolves different instructional actions by generating the subsequent skeletons, which find the most probable action to represent the partially observed action by learning the way of perceiving the tendency of the ongoing action. We conduct experiments on seven widely used action datasets: NTU-60, NTU-120, UCF101, UT-Interaction, BIT, PKU-MMD and HMDB51, and our experimental results clearly demonstrate that our method achieves very competitive performance with state-of-the-art. [ABSTRACT FROM AUTHOR]

In this paper, we propose a weakly-supervised approach called Order-Constrained Representation Learning (OCRL) to predict future actions from instructional videos by observing incomplete steps of actions. Most conventional methods focus on predicting actions based on partially observed video frames, which mainly study low-level semantics such as motion consistency. Unlike performing a single action, completing a task in an instructional video usually requires several steps of action and longer periods. Motivated by the fact that the order of action steps is key to learning task semantics, we develop a new frame of contrastive loss, called StepNCE, to integrate the shared semantic information between step order and task semantics under the framework of the memory bank-based momentum-updating algorithm. Specifically, we learn the video representations from step order-rearranged trimmed video clips based on the proposed task-consistency rule and order-consistency rule. Our StepNCE loss can be used to pre-train a video feature encoder, which is then fine-tuned to carry out the instructional video prediction task. Our approach digs deeper into the sequential logic between different action steps with respect to a certain task, which is able to promote the video understanding methods to a new semantic level. We evaluate our method on five popular instructional video and action prediction datasets: COIN, CrossTask, UT-Interaction, BIT-Interaction, and ActivityNet v1.2, and the results show that our approach gains improvements from conventional prediction methods. [ABSTRACT FROM AUTHOR]

This paper aims for the task of text-to-video retrieval, where given a query in the form of a natural-language sentence, it is asked to retrieve videos which are semantically relevant to the given query, from a great number of unlabeled videos. The success of this task depends on cross-modal representation learning that projects both videos and sentences into common spaces for semantic similarity computation. In this work, we concentrate on video representation learning, an essential component for text-to-video retrieval. Inspired by the reading strategy of humans, we propose a Reading-strategy Inspired Visual Representation Learning (RIVRL) to represent videos, which consists of two branches: a previewing branch and an intensive-reading branch. The previewing branch is designed to briefly capture the overview information of videos, while the intensive-reading branch is designed to obtain more in-depth information. Moreover, the intensive-reading branch is aware of the video overview captured by the previewing branch. Such holistic information is found to be useful for the intensive-reading branch to extract more fine-grained features. Extensive experiments on three datasets are conducted, where our model RIVRL achieves a new state-of-the-art on TGIF and VATEX. Moreover, on MSR-VTT, our model using two video features shows comparable performance to the state-of-the-art using seven video features and even outperforms models pre-trained on the large-scale HowTo100M dataset. Code is available at https://github.com/LiJiaBei-7/rivrl. [ABSTRACT FROM AUTHOR]

Detecting anomalies in videos is a fundamental issue in public security. The majority of existing deep learning methods often perform anomaly detection based on the behavior or the trajectory of a single target. However, due to the overlaps of the crowd and the low-resolution of monitoring images, the segmentation of population is hard to implement and the features cannot be learned thoroughly, which make the methods be easily disturbed by visual elements and thus may lead to false detection sometimes. To tackle these problems, we propose the influence-aware attention to learn the representative attributes of the whole crowd. Walking pedestrians can be divided into numbers of flows, and in this paper, we aim to measure the consistency of movement patterns in the same stream and the interactions between different streams. Meanwhile, great importance is given to the relation between pedestrians and the circumstance for certain anomalies occur as a result of environmental issues. Specifically, the influence-aware attention module is composed of the motion attention and the location attention, which is designed to quantify the relations in the scene from spatial and temporal aspects. For the lack of abnormal samples, we utilize a dual generator-based framework to learn interactions among normal scenes. Experimental results on six benchmarks verify the effectiveness and robustness of our proposed method. [ABSTRACT FROM AUTHOR]

In this paper, we propose a self-supervised contrastive learning method to learn video feature representations. In traditional self-supervised contrastive learning methods, constraints from anchor, positive, and negative data pairs are used to train the model. In such a case, different samplings of the same video are treated as positives, and video clips from different videos are treated as negatives. Because the spatio-temporal information is important for video representation, we set the temporal constraints more strictly by introducing intra-negative samples. In addition to samples from different videos, negative samples are extended by breaking temporal relations in video clips from the same anchor video. With the proposed Inter-Intra Contrastive (IIC) framework, we can train spatio-temporal convolutional networks to learn feature representations from videos. Strong data augmentations, residual clips, as well as head projector are utilized to construct an improved version. Three kinds of intra-negative generation functions are proposed and extensive experiments using different network backbones are conducted on benchmark datasets. Without using pre-computed optical flow data, our improved version can outperform previous IIC by a large margin, such as 19.4% (from 36.8% to 56.2%) and 5.2% (from 15.5% to 20.7%) points improvements in top-1 accuracy on UCF101 and HMDB51 datasets for video retrieval, respectively. For video recognition, over 3% points improvements can also be obtained on these two benchmark datasets. Discussions and visualizations validate that our IICv2 can capture better temporal clues and indicate the potential mechanism. [ABSTRACT FROM AUTHOR]

Recent years have witnessed the exponential increase in the demand for ultra high definition (UHD) video compression. The third generation of Audio Video Coding Standard (AVS3), which is also known as IEEE Standard 1857.10, is the latest audio and video coding standard developed by the China AVS working group. In AVS3, targeting for UHD videos, a series of efficient coding tools have been introduced, leading to the dramatical increase of computational burden. In this scenario, real-time decoding of UHD videos becomes extremely challenging. This paper presents an improved hybrid CPU + GPU accelerated framework for AVS3 decoding. In particular, the motion vector (MV) derivation process is extracted from entropy decoding threads on the CPU. Therefore, the dependency between threads is removed and the entropy decoding can be performed by multiple threads efficiently. Regarding the GPU, we design compact data structures for transform, prediction, and in-loop filtering to reduce the burden of data transmission. A flexible information buffer supporting multi-thread random writing is further created to coordinate the computation between CPU and GPU. Through asynchronous operations on the buffer, the computation of different computing units and the data transmission between them could be performed in parallel. With NVIDIA GeForce RTX 2080Ti GPU and Intel Core i7 8700K CPU, the proposed decoder achieves 151 frames per second (fps) for 4K videos and 55 fps for 8K videos in all intra configuration. In random access configuration, 218 fps and 74 fps are obtained for 4K and 8K videos, respectively. [ABSTRACT FROM AUTHOR]

The world of sports intrinsically involves fast and complex events that are difficult for coaches, trainers and players to analyze, and also for audiences to follow. In fast paced team sports such as soccer, keeping track of all the players and analyzing their performance after every match are very challenging. Current scenarios for identifying the best talents in soccer involve word-of-mouth and coaches/recruiters scouring through hours of manually annotated videos. This is a very expensive and laborious process and also biased by the nature of the recruiters. To alleviate these problems, this paper proposes an automated system that can detect, track, classify the teams of multiple players and identify the player controlling the ball in a video. The system generates three very important tactical statistics for a player: 1) duration of ball possession, 2) number of successful passes and 3) number of successful steals. This is done by training Convolutional Neural Networks (CNNs) to (a) localize and track the players on the field, (b) classify the team of a detected player, (c) identify the player controlling the ball and (d) pooling all the information extracted from (a), (b), and (c) to generate the statistics of players. To overcome the problem that the features learned from specific soccer matches do not necessarily generalize across different soccer matches, the paper proposes minimal amount of match-specific annotation and data augmentation, using a variant of Deep Convolutional Generative Adversarial Networks (DCGAN) to improve the accuracy. Experimental results and ablation studies show that the proposed approach outperforms the state-of-the-art approaches in terms of accuracy and processing speed. [ABSTRACT FROM AUTHOR]

Lots of Deepfake videos are circulating on the Internet, which not only damages the personal rights of the forged individual, but also pollutes the web environment. What’s worse, it may trigger public opinion and endanger national security. Therefore, it is urgent to fight deep forgery. Most of the current forgery detection algorithms are based on convolutional neural networks to learn the feature differences between forged and real frames from big data. In this paper, from the perspective of image generation, we simulate the forgery process based on image generation and explore possible trace of forgery. We propose a multi-scale self-texture attention Generative Network(MSTA-Net) to track the potential texture trace in image generation process and eliminate the interference of deep forgery post-processing. Firstly, a generator with encoder-decoder is to disassemble images and performed trace generation, then we merge the generated trace image and the original map, which is input into the classifier with Resnet as the backbone. Secondly, the self-texture attention mechanism(STA) is proposed as the skip connection between the encoder and the decoder, which significantly enhances the texture characteristics in the image disassembly process and assists the generation of texture trace. Finally, we propose a loss function called Prob-tuple loss restricted by classification probability to amend the generation of forgery trace directly. To verify the performance of the MSTA-Net, we design different experiments to verify the feasibility and advancement of the method. Experimental results show that the proposed method performs well on deep forged databases represented by FaceForensics++, Celeb-DF, Deeperforensics and DFDC, and some results are reaching the state-of-the-art. [ABSTRACT FROM AUTHOR]

The great success of deep learning has boosted the fast development of video quality enhancement. However, existing methods mainly focus on enhancing the objective quality of compressed video, and ignore their perceptual quality that plays a key role in determining quality of experience (QoE) of videos. In this paper, we aim at enhancing the perceptual quality of compressed video. Our main observation is that perceptual quality enhancement mostly relies on recovering the high-frequency details with fine textures. Accordingly, we propose a novel generative adversarial network (GAN) based on multi-level wavelet packet transform (WPT), which is called multi-level wavelet-based GAN+ (MW-GAN+), to exploit high-frequency details for enhancing the perceptual quality of compressed video. In MW-GAN+, we first propose a multi-level wavelet pixel-adaptive (MWP) module to extract temporal information across video frames, such that frame similarity can be utilized in recovering high-frequency details. Then, a wavelet reconstruction network, consisting of wavelet-dense residual blocks (WDRB), is developed to recover high-frequency details in a multi-level manner for enhanced frame reconstruction. Finally, we develop a 3D discriminator to encourage temporal coherence with a 3D-CNN based architecture. Experimental results demonstrate the superiority of our method over state-of-the-art methods in enhancing the perceptual quality of compressed video. Our code is available at https://github.com/IceClear/MW-GAN. [ABSTRACT FROM AUTHOR]

Weakly-supervised video object segmentation is an emerging video task to track and segment the target given a simple bounding box label, which requires the method to fully catch and utilize the target information. Most existing approaches only rely on the guidance of a single frame and ignore the interaction between different frames when gathering information, making them hard to achieve reliable target representation. In this paper, we propose to capture the temporal dependencies and gather information from multiple frames through bilateral temporal re-aggregation. We explore three schemes to build the aggregation: 1) a two-stage re-aggregation mechanism is applied to provide target prior to the current frame, which obtains more valid feature matching and information aggregation; 2) a query-memory bilateral aggregation module is proposed to aggregate features from an unlimited amount of past frames and enable the mutual perception between different frames to validate the gathered information; 3) we guide the learning of aggregation modules through a novel cross-task representation distillation, transferring the knowledge from a semi-supervised model to our weakly-supervised model without increasing the inference latency. These schemes collaboratively build an efficient and competent aggregation process, thus we can fully exploit the video context to make the inference. Experimental results on four benchmarks show that our method achieves superior performance than previous methods and still maintains the efficiency ($e.g$., overall scores of 70.4% and 72.5% on the YouTube-VOS and DAVIS 2017 validation sets, respectively). [ABSTRACT FROM AUTHOR]

In recent years, high-definition restoration of early videos have received much attention. Real-world interlaced videos usually contain various degradations mixed with interlacing artifacts, such as noises and compression artifacts. Unfortunately, traditional deinterlacing methods only focus on the inverse process of interlacing scanning, and cannot remove these complex and complicated artifacts. Hence, this paper proposes an image deinterlacing network (DIN), which is specifically designed for joint removal of interlacing mixed with other artifacts. The DIN is composed of two stages, i.e., a cooperative vertical interpolation stage for splitting and fully using the information of adjacent fields, and a field-merging stage to perceive movements and suppress ghost artifacts. Experimental results demonstrate the effectiveness of the proposed DIN on both synthetic and real-world test sets. [ABSTRACT FROM AUTHOR]

Video inpainting is a task of synthesizing spatio-temporal coherent content in missing regions of the given video sequence, which has recently drawn increasing attention. To utilize the temporal information across frames, most recent deep learning-based methods align reference frames to target frame firstly with explicit or implicit motion estimation and then integrate the information from the aligned frames. However, their performance relies heavily on the accuracy of frame-to-frame alignment. To alleviate the above problem, in this paper, a novel Temporal Group Fusion Network (TGF-Net) is proposed to effectively integrate temporal information through a two-stage fusion strategy. Specifically, the input frames are reorganized into different groups, where each group is followed by an intra-group fusion module to integrate information within the group. Different groups provide complementary information for the missing region. A temporal attention model is further designed to adaptively integrate the information across groups. Such a temporal information fusion way gets rid of the dependence on alignment operations, greatly improving the visual quality and temporal consistency of the inpainted results. In addition, a coarse alignment model is introduced at the beginning of the network to handle videos with large motion. Extensive experiments on DAVIS and Youtube-VOS datasets demonstrate the superiority of our proposed method in terms of PSNR/SSIM values, visual quality and temporal consistency, respectively. [ABSTRACT FROM AUTHOR]

With the evolution of electronic devices, such as 3D cameras, addressing the challenges of text localization in 3D video (e.g., for indexing) is increasingly drawing the attention of the multimedia and video processing community. Existing methods focus on 2D video and their performance in the presence of the challenges in 3D video, such as shadow areas associated with text and irregularly sized and shaped text, degrades. This paper proposes the first approach that successfully addresses the challenges of 3D video in addition to those of 2D. It employs a number of innovations, among which, the first is the Generalized Gradient Vector Flow (GGVF) for dominant points detection. The second is the Wavefront concept for text candidate point detection from those dominant points. In addition, an Adaptive B-Spline Polygon Curve Network (ABS-Net) is proposed for accurate text localization in 3D videos by constructing tight fitting bounding polygons using text candidate points. Extensive experiments on custom (3D video) and standard datasets (2D video and scene text) show that the proposed method is practical and useful, and overall outperforms existing state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Self-supervised representation learning for videos has been very attractive recently because these methods exploit the information inherently obtained from the video itself instead of annotated labels that is quite time-consuming. However, existing methods ignore the importance of global observation while performing spatio-temporal transformation perception, which highly limits the expression capabilities of the video representation. This paper proposes a novel pretext task that combines the temporal information perception of the video with the motion amplitude perception of moving objects to learn the spatio-temporal representation of the video. Specifically, given a video clip containing several video segments, each video segment is sampled by different sampling rates and the order of video segments is disrupted. Then, the network is used to regress the sampling rate of each video segment and classify the order of input video segments. In the pre-training stage, the network can learn rich spatio-temporal semantic information where content-related contrastive learning is introduced to make the learned video representation more discriminative. To alleviate the appearance dependency caused by contrastive learning, we design a novel and robust vector similarity measurement approach, which can take feature alignment into consideration. Moreover, a view synthesis framework is proposed to further improve the performance of contrastive learning by automatically generating reasonable transformed views. We conduct benchmark experiments with several 3D backbone networks on two datasets. The results show that our proposed method outperforms the existing state-of-the-art methods across the three backbones on two downstream tasks of human action recognition and video retrieval. [ABSTRACT FROM AUTHOR]

Video highlight detection (VHD) has emerged as a pressing task due to the unprecedentedly increasing amount of video data, such as those from e-commerce live-broadcasting platforms. Many approaches focus on exploiting text data, in the form of video description or time-sync comments, to facilitate the VHD task. Despite the promising results, they have largely overlooked the noises inherent in the text data and have mostly relied on isolating the feature of text and video. In this paper, we introduce a novel model to handle VHD, termed Coevolution Network (CoEvo-Net), that allows us to account for joint learning of the language and video features explicitly via a coevolution paradigm, in which features from the two data modalities progressively refine each other. This is achieved by a dedicated CoEvo-Cell that takes language and video together as inputs, extracts cross-modality, and filters the undesired parts of the input, such as words in a sentence. Furthermore, we release a large-scale dataset of e-commerce for VHD, in which each video is coupled with a sentence for description, to benchmark the sentence-based VHD approaches. Extensive experiments on the released dataset demonstrate that CoEvo-Net achieves state-of-the-art performance. Our dataset and code will be made publicly available. [ABSTRACT FROM AUTHOR]

Face manipulation techniques improve fast with the development of powerful image generation models. Two particular face manipulation methods, namely face swap and expression reenactment attract much attention for their flexibility and ease to generate high quality synthesis results. Recently, these two subjects are actively studied. However, most existing methods treat the two tasks separately, ignoring their underlying similarity. In this paper, we propose to tackle the two problems within a unified framework that achieves high quality synthesis results. The enabling component for our unified framework is the clean disentanglement of 3D pose, shape, and expression factors and then recombining them for different tasks accordingly. We then use the same set of 2D representations for face swap and expression reenactment tasks that are input to a common image translation model to directly generate the final synthetic images. Once trained, the proposed model can accomplish both face swap and expression reenactment tasks for previously unseen subjects. Comprehensive experiments and comparisons show that the proposed method achieves high fidelity results in multiple aspects, and it is especially good at faithfully preserving source facial shape in the face swap task, and accurately transferring facial movements in the expression reenactment task. [ABSTRACT FROM AUTHOR]

The emergence of wearable devices has opened up new potentials for egocentric activity recognition. Although some methods integrate attention mechanisms into deep neural networks to capture fine-grained human-object interactions in a weak-supervision manner, they either ignore exploiting the temporal consistency or generate attention based on considering appearance cues only. To address these limitations, in this paper, we propose an enhanced attention-tracking method, combined with multi-branch network (EAT-MBNet), for egocentric activity recognition. Specifically, we propose class-aware attention maps (CAAMs) by employing a self-attention-based module to refine the class activation maps (CAMs). Our proposed method can enhance the semantic dependency between the activity categories and the feature maps. To highlight the discriminative features from the regions of interest across frames, we propose a flow-guided attention-tracking (F-AT) module, by simultaneously leveraging historical attention and motion patterns. Furthermore, we propose a cross-modality modeling branch based on an interactive GRU module, which captures the time-synchronized long-term relationships between the appearance and motion branches. Experimental results on four egocentric activity benchmarks demonstrate that the proposed method achieves state-of-the-art performance. [ABSTRACT FROM AUTHOR]

Recently, VR sickness assessment for VR videos is highly demanded in industry and research fields to address VR viewing safety issues. Especially, it is difficult to evaluate VR sickness of individuals due to individual differences. To achieve the challenging goal, we focus on deep feature fusion of sickness-related information. In this paper, we propose a novel deep learning-based assessment framework which estimates VR sickness of individual viewers with VR videos and corresponding physiological responses. We design the content stimulus guider imitating the phenomenon that humans feel VR sickness. The content stimulus guider extracts a deep stimulus feature from a VR video to reflect VR sickness caused by VR videos. In addition, we devise the physiological response guider to encode physiological responses that are acquired while humans experience VR videos. Each physiology sickness feature extractor (EEG, ECG, and GSR) in the physiological response guider is designed to suit their physiological characteristics. Extracted physiology sickness features are then fused into a deep physiology feature that comprehensively reflects individual deviations of VR sickness. Finally, the VR sickness predictor assesses individual VR sickness effectively with the fusion of the deep stimulus feature and the deep physiology feature. To validate the proposed method extensively, we built two benchmark datasets which contain 360-degree VR videos with physiological responses (EEG, ECG, and GSR) and SSQ scores. Experimental results show that the proposed method achieves meaningful correlations with human SSQ scores. Further, we validate the effectiveness of the proposed network designs by conducting analysis on feature fusion and visualization. [ABSTRACT FROM AUTHOR]

During the past few years, face videos, e.g., video conference, interviews and variety shows, have grown explosively with millions of users over social media networks. Unfortunately, the existing compression algorithms are applied to these videos for reducing bandwidth, which also bring annoying artifacts to face regions. This paper addresses the problem of face quality enhancement in compressed videos by reducing the artifacts of face regions. Specifically, we establish a compressed face video (CFV) database, which includes 196,337 faces in 214 high-quality video sequences and their corresponding 1,712 compressed sequences. We find that the faces of compressed videos exhibit tremendous scale variation and quality fluctuation. Motivated by scalable video coding, we propose a multi-scale recurrent scalable network (MRS-Net+) to enhance the quality of multi-scale faces in compressed videos. The MRS-Net+ is comprised by one base and two refined enhancement levels, corresponding to the quality enhancement of small-, medium- and large-scale faces, respectively. In the multi-level architecture of our MRS-Net+, small-/medium-scale face quality enhancement serves as the basis for facilitating the quality enhancement of medium-/large-scale faces. We further develop a landmark-assisted pyramid alignment (LPA) subnet to align faces across consecutive frames, and then apply the mask-guided quality enhancement (QE) subnet for enhancing multi-scale faces. Finally, experimental results show that our MRS-Net+ method achieves averagely 1.196 dB improvement of peak signal-to-noise ratio (PSNR) and 23.54% saving of Bjøntegaard distortion-rate (BD-rate), significantly outperforming other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Previous video salient object detection (VSOD) approaches have mainly focused on the perspective of network design for achieving performance improvements. However, with the recent slowdown in the development of deep learning techniques, it might become increasingly difficult to anticipate another breakthrough solely via complex networks. Therefore, this paper proposes a universal learning scheme to obtain a further 3% performance improvement for all state-of-the-art (SOTA) VSOD models. The major highlight of our method is that we propose the ‘motion quality’, a new concept for mining video frames from the ‘buffered’ testing video stream for constructing a fine-tuning set. By using our approach, all frames in this set can all well-detect their salient object by the ‘target SOTA model’ — the one we want to improve. Thus, the VSOD results of the mined set, which were previously derived by the target SOTA model, can be directly applied as pseudolearning objectives to fine-tune a completely new spatial model that has been pretrained on the widely used DAVIS-TR set. Since some spatial scenes in the buffered testing video stream are shown, the fine-tuned spatial model can perform very well for the remaining unseen testing frames, outperforming the target SOTA model significantly. Although offline model fine tuning requires additional time costs, the performance gain can still benefit scenarios without speed requirements. Moreover, its semisupervised methodology might have considerable potential to inspire the VSOD community in the future. [ABSTRACT FROM AUTHOR]

In this paper, we propose a Structure-adaptive Neighborhood Preserving Hashing (SNPH) method for unsupervised scalable video search. Unlike most existing hashing methods which equally encode an entire video into a binary feature vector, we propose a neighborhood attention mechanism which encodes the neighborhood-relevant content of a video to better preserve the neighborhood relationships among videos. Motivated by the fact that a video usually contains multiple shots and each shot depicts a different activity, we further develop a structure-adaptive encoder to model the hierarchical structure of the video. Specifically, the encoder adaptively divides each video into multiple segments via detecting temporal boundaries across frames and encodes these segments as a compact binary vector to capture rich structural information. We integrate the neighborhood attention mechanism into the structure-adaptive encoder to learn hash functions that jointly preserve the neighborhood relationships among videos and exploit the hierarchical structure in a video. Experimental results on three widely used benchmark datasets show that our proposed method consistently outperforms state-of-the-art unsupervised video hashing methods. [ABSTRACT FROM AUTHOR]

Due to the proliferation of sports-related multimedia content on the WWW, effective visual search and retrieval present interesting research challenges. These are caused by poor image quality, a wide range of possible camera points of view, pose variations on the part of athletes engaged in playing a sport, deformations of text appearing on sports person’s clothing and uniforms in motion, occlusions caused by other objects, etc. To address these challenges, this paper presents a new method for detecting text on human bodies in sports images. Unlike most existing methods, which attempt to exploit locations of a player’s torso, face, and skin, we propose an end-to-end episodic learning approach that employs inductive learning criteria for detecting clothing regions in an image, which are, in turn, then used for text detection. Our method integrates a Residual Network (ResNet) and Pyramidal Pooling Module (PPM) for generating a spatial attention map. The Progressive Scalable Expansion Algorithm (PSE) is adapted for text detection from these regions. Experimental results on our own dataset as well as several benchmarks (like RBNR and MMM which contain images of runners in marathons, and Re-ID which is a person re-identification dataset) demonstrate that the proposed method outperforms existing methods in terms of precision and F1-score. We also present results for sports images chosen from natural scene text detection datasets such as CTW1500 and MS-COCO to show the proposed method is effective and reliable across a range of inputs. [ABSTRACT FROM AUTHOR]

Recently, many researches on convolution neural network (CNN) based in-loop filters have been proposed to improve coding efficiency. However, most existing CNN based filters tend to train and deploy multiple networks for various quantization parameters (QP) and frame types (FT), which drastically increases resources in training these models and the memory burdens for video codec. In this paper, we propose a novel variable CNN (VCNN) based in-loop filter for VVC, which can effectively handle the compressed videos with different QPs and FTs via a single model. Specifically, an efficient and flexible attention module is developed to recalibrate features according to QPs or FTs. Then we embed the module into the residual block so that these informative features can be continuously utilized in the residual learning process. To minimize the information loss in the learning process of the entire network, we utilize a residual feature aggregation module (RFA) for more efficient feature extraction. Based on it, an efficient network architecture VCNN is designed that can not only effectively reduce compression artifacts, but also can be adaptive to various QPs and FTs. To address training data imbalance on various QPs and FTs and improve the robustness of the model, a focal mean square error loss function is employed to train the proposed network. Then we integrate the VCNN into VVC as an additional tool of in-loop filters after the deblocking filter. Extensive experimental results show that our VCNN approach obtains on average 3.63%, 4.36%, 4.23%, 3.56% under all intra, low-delay P, low-delay, and random access configurations, respectively, which is even better than QP-Separate models. [ABSTRACT FROM AUTHOR]

Facial expression recognition is important for human–computer interaction and other applications. Several facial expression datasets have been published in recent decades and have enabled improvements in algorithms for classifying emotions. However, recognition of realistic expressions in real-world conditions is still challenging because of uncontrolled conditions, such as lighting, brightness, pose, and occlusion. In this paper, we propose a light attention embedding network based on the spatial attention mechanism (LAENet-SA), which can focus on locations in an image that are relevant to emotion. LAENet-SA allows a small number of attention modules to be embedded and can be constructed from typical convolutional neural networks. The performance of LAENet-SA on facial expression recognition has been validated on three facial expression datasets, including a lab-controlled dataset and two in-the-wild datasets. Experimental results show that LAENet-SA improved the performance on each dataset, compared with state-of-the-art methods, and achieved better generalization when tested on facial images with occlusion. [ABSTRACT FROM AUTHOR]

The attribution method provides a direction for interpreting opaque neural networks in a visual way by identifying and visualizing the input regions/pixels that dominate the output of a network. Regarding the attribution method for visually explaining video understanding networks, it is challenging because of the unique spatiotemporal dependencies existing in video inputs and the special 3D convolutional or recurrent structures of video understanding networks. However, most existing attribution methods focus on explaining networks taking a single image as input and a few works specifically devised for video attribution come short of dealing with diversified structures of video understanding networks. In this paper, we investigate a generic perturbation-based attribution method that is compatible with diversified video understanding networks. Besides, we propose a novel regularization term to enhance the method by constraining the smoothness of its attribution results in both spatial and temporal dimensions. In order to assess the effectiveness of different video attribution methods without relying on manual judgement, we introduce reliable objective metrics which are checked by a newly proposed reliability measurement. We verified the effectiveness of our method by both subjective and objective evaluation and comparison with multiple significant attribution methods. [ABSTRACT FROM AUTHOR]

3D human trajectory reconstruction in an indoor scene is critical in various applications, such as indoor navigation and human activity recognition. This task is challenging due to occlusion and clutters of indoor scenes, flexible human body joints, and severe lack of relevant datasets. Although several methods have been proposed to reconstruct a 3D human trajectory, they either can recover only 2D positions or require human initiative cooperation. In this paper, we propose a novel framework for 3D human trajectory reconstruction in an indoor scene using monocular surveillance videos and static point clouds without any initiative cooperation. The proposed framework consists of three modules: 3D pose estimation, depth regression, and trajectory reconstruction. We first estimate 3D pose from videos. Especially, we reconstruct a half-body 3D pose to deal with the occlusion problem. Then, we propose a depth regression approach to iteratively regress the depth of a 3D pose. Unlike data-driven approaches, our depth regression approach does not require training data and can be integrated into any 3D pose model. Finally, we exploit the geometric constraints from the point cloud to optimize the 3D trajectory. We evaluated the 3D pose estimation and depth regression modules on the H3.6M datasets. Due to the lack of evaluation datasets, we also built a trajectory dataset to evaluate the trajectory reconstruction performance. Empirical evaluation shows that our framework achieves accurate trajectory reconstruction results on real-world videos. [ABSTRACT FROM AUTHOR]

In this paper, we are concerned with the detection of progressive dynamic saliency from video sequences. More precisely, we are interested in saliency related to motion and likely to appear progressively over time. It can be relevant to trigger alarms, to dedicate additional processing or to detect specific events. Trajectories represent the best way to support progressive dynamic saliency detection. Accordingly, we will talk about trajectory saliency. A trajectory will be qualified as salient if it deviates from normal trajectories that share a common motion pattern related to a given context. First, we need a compact while discriminative representation of trajectories. We adopt a (nearly) unsupervised learning-based approach. The latent code estimated by a recurrent auto-encoder provides the desired representation. In addition, we enforce consistency for normal (similar) trajectories through the auto-encoder loss function. The distance of the trajectory code to a prototype code accounting for normality is the means to detect salient trajectories. We validate our trajectory saliency detection method on synthetic and real trajectory datasets, and highlight the contributions of its different components. We compare our method favourably to existing methods on several saliency configurations constructed from the publicly available large dataset of pedestrian trajectories acquired in a railway station. [ABSTRACT FROM AUTHOR]

Passive video forensics has become an active topic in recent years. Generally, a pristine video obtained from surveillance cameras or other video recording devices is single compressed. At the same time, double lossy compression will certainly be introduced in tampered videos since it needs to go through recompression to perform tampering on a commonly used compressed video. In the video’s double compression, some traces are left due to the intrinsic effects of recompression. Traditional supervised learning is inefficient because it requires two classes (the pristine and the manipulated) that occur in the video to be exhaustively assigned labels. Actually, compared with pristine videos, manipulated videos with labels are more difficult to obtain. To address this problem, in this paper, one-class classification, which is often used for anomaly detection and only needs the target class, is introduced. We first treat all decompressed video frames as still images and extract subtractive pixel adjacency matrix (SPAM) steganalysis features to detect traces left in the double compression process. Then we adopt a Gaussian density-based one-class classifier since SPAM features extracted from pristine video frames approximately subject to the Gaussian distribution. Furthermore, we improve the robustness of the classifier by using ensemble strategy. Experimental results indicate that our proposed method exceeds other more complex one-class classification methods, and outperforms fully-supervised learning methods only by feeding features from single compressed video frames to our one-class classifier. [ABSTRACT FROM AUTHOR]

Automatically synthesizing sounds for different visual contents poses a challenge and there is a strong need to facilitate the direct creation of realistic sounds. Different from previous works, in this paper, we propose a novel deep learning based approach, which formulates sound simulation as a regression problem. This allows us to circumvent the complexity of the acoustic theory by a novel, general-purpose neural sound synthesis (V2RA) network. Moreover, the end-to-end architecture of V2RA ensures full training without any extra inputs, which thereby greatly improves the scalability and reusability over previous works. In contrast to conventional visual-to-audio generation methods, the V2RA problem is established and solved by generative adversarial networks (GANs). Furthermore, our network architecture can directly predict synchronized raw audio signals (unlike most existing approaches that handle the audio through spectrograms) and generate sound in real time. To evaluate the performance of the neural network generator, we specifically introduce two quantitative scores. Various experiments demonstrate that our V2RA network can produce compelling sound results, which thus provides a viable solution for applications such as sound design and dubbing. [ABSTRACT FROM AUTHOR]

Human visual system has a limitation of sensitivity in detecting small distortion in an image/video and the minimum perceptual threshold is so called Just Noticeable Difference (JND). JND modelling is challenging since it highly depends on visual contents and perceptual factors are not fully understood. In this paper, we propose deep learning based JND and perceptual quality prediction models, which are able to predict the Satisfied User Ratio (SUR) and Video Wise JND (VWJND) of compressed videos with different resolutions and coding parameters. Firstly, the SUR prediction is modeled as a regression problem that fits deep learning tools. Then, Video Wise Spatial SUR method (VW-SSUR) is proposed to predict the SUR value for compressed video, which mainly considers the spatial distortion. Thirdly, we further propose Video Wise Spatial-Temporal SUR (VW-STSUR) method to improve the SUR prediction accuracy by considering the spatial and temporal information. Two fusion schemes that fuse the spatial and temporal information in quality score level and in feature level, respectively, are investigated. Finally, key factors including key frame and patch selections, cross resolution prediction and complexity are analyzed. Experimental results demonstrate the proposed VW-SSUR method outperforms in both SUR and VWJND prediction as compared with the state-of-the-art schemes. Moreover, the proposed VW-STSUR further improves the accuracy as compared with the VW-SSUR and the conventional JND models, where the mean SUR prediction error is 0.049, and mean VWJND prediction error is 1.69 in quantization parameter and 0.84 dB in peak signal-to-noise ratio. [ABSTRACT FROM AUTHOR]

The development of technologies that can generate Deepfake videos is expanding rapidly. These videos are easily synthesized without leaving obvious traces of manipulation. Though forensically detection in high-definition video datasets has achieved remarkable results, the forensics of compressed videos is worth further exploring. In fact, compressed videos are common in social networks, such as videos from Instagram, Wechat, and Tiktok. Therefore, how to identify compressed Deepfake videos becomes a fundamental issue. In this paper, we propose a two-stream method by analyzing the frame-level and temporality-level of compressed Deepfake videos. Since the video compression brings lots of redundant information to frames, the proposed frame-level stream gradually prunes the network to prevent the model from fitting the compression noise. Aiming at the problem that the temporal consistency in Deepfake videos might be ignored, we apply a temporality-level stream to extract temporal correlation features. When combined with scores from the two streams, our proposed method performs better than the state-of-the-art methods in compressed Deepfake videos detection. [ABSTRACT FROM AUTHOR]

Multimodal emotion recognition in conversational videos (ERC) develops rapidly in recent years. To fully extract the relative context from video clips, most studies build their models on the entire dialogues which make them lack of real-time ERC ability. Different from related researches, a novel multimodal emotion recognition model for conversational videos based on reinforcement learning and domain knowledge (ERLDK) is proposed in this paper. In ERLDK, the reinforcement learning algorithm is introduced to conduct real-time ERC with the occurrence of conversations. The collection of history utterances is composed as an emotion-pair which represents the multimodal context of the following utterance to be recognized. Dueling deep-Q-network (DDQN) based on gated recurrent unit (GRU) layers is designed to learn the correct action from the alternative emotion categories. Domain knowledge is extracted from public dataset based on the former information of emotion-pairs. The extracted domain knowledge is used to revise the results from the RL module and is transformed into other dataset to examine the rationality. The experimental results on datasets show that ERLDK achieves the state-of-the-art results on weighted average and most of the specific emotion categories. [ABSTRACT FROM AUTHOR]