Search

Effectively evaluating the perceptual quality of dehazed images remains an under-explored research issue. In this paper, we propose a no-reference complex-valued convolutional neural network (CV-CNN) model to conduct automatic dehazed image quality evaluation. Specifically, a novel CV-CNN is employed that exploits the advantages of complex-valued representations, achieving better generalization capability on perceptual feature learning than real-valued ones. To learn more discriminative features to analyze the perceptual quality of dehazed images, we design a dual-stream CV-CNN architecture. The dual-stream model comprises a distortion-sensitive stream that operates on the dehazed RGB image, and a haze-aware stream on a novel dark channel difference image. The distortion-sensitive stream accounts for perceptual distortion artifacts, while the haze-aware stream addresses the possible presence of residual haze. Experimental results on three publicly available dehazed image quality assessment (DQA) databases demonstrate the effectiveness and generalization of our proposed CV-CNN DQA model as compared to state-of-the-art no-reference image quality assessment algorithms. [ABSTRACT FROM AUTHOR]

Most researchers focus on designing accurate crowd counting models with heavy parameters and computations but ignore the resource burden during the model deployment. A real-world scenario demands an efficient counting model with low-latency and high-performance. Knowledge distillation provides an elegant way to transfer knowledge from a complicated teacher model to a compact student model while maintaining accuracy. However, the student model receives the wrong guidance with the supervision of the teacher model due to the inaccurate information understood by the teacher in some cases. In this paper, we propose a dual-knowledge distillation (DKD) framework, which aims to reduce the side effects of the teacher model and transfer hierarchical knowledge to obtain a more efficient counting model. First, the student model is initialized with global information transferred by the teacher model via adaptive perspectives. Then, the self-knowledge distillation forces the student model to learn the knowledge by itself, based on intermediate feature maps and target map. Specifically, the optimal transport distance is utilized to measure the difference of feature maps between the teacher and the student to perform the distribution alignment of the counting area. Extensive experiments are conducted on four challenging datasets, demonstrating the superiority of DKD. When there are only approximately 6% of the parameters and computations from the original models, the student model achieves a faster and more accurate counting performance as the teacher model even surpasses it. [ABSTRACT FROM AUTHOR]

Efficient action recognition aims to classify a video clip into a specific action category with a low computational cost. It is challenging since the integrated spatial-temporal calculation (e. g., 3D convolution) introduces intensive operations and increases complexity. This paper explores the feasibility of the integration of channel splitting and filter decoupling for efficient architecture design and feature refinement by proposing a novel spatio-temporal collaborative (STC) module. STC splits the video feature channels into two groups and separately learns spatio-temporal representations in parallel with decoupled convolutional operators. Particularly, STC consists of two computation-efficient blocks, i.e., $\text {S}_{\mathrm{ T}}$ and $\text {T}_{\mathrm{ S}}$ , where they extract either spatial (${S}_{\cdot }$) or temporal (${T}_{\cdot }$) features and further refine their features with either temporal ($\cdot _{T}$) or spatial ($\cdot _{S}$) contexts globally. The spatial/temporal context refers to information dynamics aggregated from temporal/spatial axis. To thoroughly examine our method’s performance in video action recognition tasks, we conduct extensive experiments using five video benchmark datasets requiring temporal reasoning. Experimental results show that the proposed STC networks achieve a competitive trade-off between model efficiency and effectiveness. [ABSTRACT FROM AUTHOR]

Many unsupervised domain adaptation (UDA) methods have been developed and have achieved promising results in various pattern recognition tasks. However, most existing methods assume that raw source data are available in the target domain when transferring knowledge from the source to the target domain. Due to the emerging regulations on data privacy, the availability of source data cannot be guaranteed when applying UDA methods in a new domain. The lack of source data makes UDA more challenging, and most existing methods are no longer applicable. To handle this issue, this paper analyzes the cross-domain representations in source-data-free unsupervised domain adaptation (SF-UDA). A new theorem is derived to bound the target-domain prediction error using the trained source model instead of the source data. On the basis of the proposed theorem, information bottleneck theory is introduced to minimize the generalization upper bound of the target-domain prediction error, thereby achieving domain adaptation. The minimization is implemented in a variational inference framework using a newly developed latent alignment variational autoencoder (LA-VAE). The experimental results show good performance of the proposed method in several cross-dataset classification tasks without using source data. Ablation studies and feature visualization also validate the effectiveness of our method in SF-UDA. [ABSTRACT FROM AUTHOR]

Low-rank tensor representation philosophy has enjoyed a reputation in many hyperspectral image (HSI) low-level vision applications, but previous studies often failed to comprehensively exploit the low-rank nature of HSI along different modes in low-dimensional subspace, and unsurprisingly handled only one specific task. To address these challenges, in this paper, we figured out that in addition to the spatial correlation, the spectral dependency of HSI also implicitly exists in the coefficient tensor of its subspace, this crucial dependency that was not fully utilized by previous studies yet can be effectively exploited in a cascaded manner. This led us to propose a unified subspace low-rank learning regime with a new tensor cascaded rank minimization, named STCR, to fully couple the low-rankness of HSI in different domains for various low-level vision tasks. Technically, the high-dimensional HSI was first projected into a low-dimensional tensor subspace, then a novel tensor low-cascaded-rank decomposition was designed to collapse the constructed tensor into three core tensors in succession to more thoroughly exploit the correlations in spatial, nonlocal, and spectral modes of the coefficient tensor. Next, difference continuity-regularization was introduced to learn a basis that more closely approximates the HSI’s endmembers. The proposed regime realizes a comprehensive delineation of the self-portrait of HSI tensor. Extensive evaluations conducted with dozens of state-of-the-art (SOTA) baselines on eight datasets verified that the proposed regime is highly effective and robust to typical HSI low-level vision tasks, including denoising, compressive sensing reconstruction, inpainting, and destriping. The source code of our method is released at https://github.com/CX-He/STCR.git. [ABSTRACT FROM AUTHOR]

Recently, graph-based methods have been widely applied to model fitting. However, in these methods, association information is invariably lost when data points and model hypotheses are mapped to the graph domain. In this paper, we propose a novel model fitting method based on co-clustering on bipartite graphs (CBG) to estimate multiple model instances in data contaminated with outliers and noise. Model fitting is reformulated as a bipartite graph partition behavior. Specifically, we use a bipartite graph reduction technique to eliminate some insignificant vertices (outliers and invalid model hypotheses), thereby improving the reliability of the constructed bipartite graph and reducing the computational complexity. We then use a co-clustering algorithm to learn a structured optimal bipartite graph with exact connected components for partitioning that can directly estimate the model instances (i.e., post-processing steps are not required). The proposed method fully utilizes the duality of data points and model hypotheses on bipartite graphs, leading to superior fitting performance. Exhaustive experiments show that the proposed CBG method performs favorably when compared with several state-of-the-art fitting methods. [ABSTRACT FROM AUTHOR]

Existing crowd counting designs usually exploit multi-branch structures to address the scale diversity problem. However, branches in these structures work in a competitive rather than collaborative way. In this paper, we focus on promoting collaboration between branches. Specifically, we propose an attention-guided collaborative counting module (AGCCM) comprising an attention-guided module (AGM) and a collaborative counting module (CCM). The CCM promotes collaboration among branches by recombining each branch’s output into an independent count and joint counts with other branches. The AGM capturing the global attention map through a transformer structure with a pair of foreground-background related loss functions can distinguish the advantages of different branches. The loss functions do not require additional labels and crowd division. In addition, we design two kinds of bidirectional transformers (Bi-Transformers) to decouple the global attention to row attention and column attention. The proposed Bi-Transformers are able to reduce the computational complexity and handle images in any resolution without cropping the image into small patches. Extensive experiments on several public datasets demonstrate that the proposed algorithm performs favorably against the state-of-the-art crowd counting methods. [ABSTRACT FROM AUTHOR]

Skeleton-based action recognition is widely used in varied areas, e.g., surveillance and human-machine interaction. Existing models are mainly learned in a supervised manner, thus heavily depending on large-scale labeled data, which could be infeasible when labels are prohibitively expensive. In this paper, we propose a novel Contrast-Reconstruction Representation Learning network (CRRL) that simultaneously captures postures and motion dynamics for unsupervised skeleton-based action recognition. It consists of three parts: Sequence Reconstructor (SER), Contrastive Motion Learner (CML), and Information Fuser (INF). SER learns representation from skeleton coordinate sequence via reconstruction. However the learned representation tends to focus on trivial postural coordinates and be hesitant in motion learning. To enhance the learning of motions, CML performs contrastive learning between the representation learned from coordinate sequences and additional velocity sequences, respectively. Finally, in the INF module, we explore varied strategies to combine SER and CML, and propose to couple postures and motions via a knowledge-distillation based fusion strategy which transfers the motion learning from CML to SER. Experimental results on several benchmarks, i.e., NTU RGB+D 60/120, PKU-MMD, CMU, and NW-UCLA, demonstrate the promise of the our method by outperforming state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

Domain adaptation is to transfer the shared knowledge learned from the source domain to a new environment, i.e., target domain. One common practice is to train the model on both labeled source-domain data and unlabeled target-domain data. Yet the learned models are usually biased due to the strong supervision of the source domain. Most researchers adopt the early-stopping strategy to prevent over-fitting, but when to stop training remains a challenging problem since the lack of the target-domain validation set. In this paper, we propose one efficient bootstrapping method, called Adaboost Student, explicitly learning complementary models during training and liberating users from empirical early stopping. Adaboost Student combines deep model learning with the conventional training strategy, i.e., adaptive boosting, and enables interactions between learned models and the data sampler. We adopt one adaptive data sampler to progressively facilitate learning on hard samples and aggregate “weak” models to prevent over-fitting. Extensive experiments show that (1) Without the need to worry about the stopping time, AdaBoost Student provides one robust solution by efficient complementary model learning during training. (2) AdaBoost Student is orthogonal to most domain adaptation methods, which can be combined with existing approaches to further improve the state-of-the-art performance. We have achieved competitive results on three widely-used scene segmentation domain adaptation benchmarks. [ABSTRACT FROM AUTHOR]

Video captioning targets interpreting the complex visual contents as text descriptions, which requires the model to fully understand video scenes including objects and their interactions. Prevailing methods adopt off-the-shelf object detection networks to give object proposals and use the attention mechanism to model the relations between objects. They often miss some undefined semantic concepts of the pretrained model and fail to identify exact predicate relationships between objects. In this paper, we investigate an open research task of generating text descriptions for the given videos, and propose Cross-Modal Graph (CMG) with meta concepts for video captioning. Specifically, to cover the useful semantic concepts in video captions, we weakly learn the corresponding visual regions for text descriptions, where the associated visual regions and textual words are named cross-modal meta concepts. We further build meta concept graphs dynamically with the learned cross-modal meta concepts. We also construct holistic video-level and local frame-level video graphs with the predicted predicates to model video sequence structures. We validate the efficacy of our proposed techniques with extensive experiments and achieve state-of-the-art results on two public datasets. [ABSTRACT FROM AUTHOR]

In this paper, we address the Online Unsupervised Domain Adaptation (OUDA) problem and propose a novel multi-stage framework to solve real-world situations when the target data are unlabeled and arriving online sequentially in batches. Most of the traditional manifold-based methods on the OUDA problem focus on transforming each arriving target data to the source domain without sufficiently considering the temporal coherency and accumulative statistics among the arriving target data. In order to project the data from the source and the target domains to a common subspace and manipulate the projected data in real-time, our proposed framework institutes a novel method, called an Incremental Computation of Mean-Subspace (ICMS) technique, which computes an approximation of mean-target subspace on a Grassmann manifold and is proven to be a close approximate to the Karcher mean. Furthermore, the transformation matrix computed from the mean-target subspace is applied to the next target data in the recursive-feedback stage, aligning the target data closer to the source domain. The computation of transformation matrix and the prediction of next-target subspace leverage the performance of the recursive-feedback stage by considering the cumulative temporal dependency among the flow of the target subspace on the Grassmann manifold. The labels of the transformed target data are predicted by the pre-trained source classifier, then the classifier is updated by the transformed data and predicted labels. Extensive experiments on six datasets were conducted to investigate in depth the effect and contribution of each stage in our proposed framework and its performance over previous approaches in terms of classification accuracy and computational speed. In addition, the experiments on traditional manifold-based learning models and neural-network-based learning models demonstrated the applicability of our proposed framework for various types of learning models. [ABSTRACT FROM AUTHOR]

We consider the problem of obtaining image quality representations in a self-supervised manner. We use prediction of distortion type and degree as an auxiliary task to learn features from an unlabeled image dataset containing a mixture of synthetic and realistic distortions. We then train a deep Convolutional Neural Network (CNN) using a contrastive pairwise objective to solve the auxiliary problem. We refer to the proposed training framework and resulting deep IQA model as the CONTRastive Image QUality Evaluator (CONTRIQUE). During evaluation, the CNN weights are frozen and a linear regressor maps the learned representations to quality scores in a No-Reference (NR) setting. We show through extensive experiments that CONTRIQUE achieves competitive performance when compared to state-of-the-art NR image quality models, even without any additional fine-tuning of the CNN backbone. The learned representations are highly robust and generalize well across images afflicted by either synthetic or authentic distortions. Our results suggest that powerful quality representations with perceptual relevance can be obtained without requiring large labeled subjective image quality datasets. The implementations used in this paper are available at https://github.com/pavancm/CONTRIQUE. [ABSTRACT FROM AUTHOR]

Point cloud (PC) is a collection of discrete geometric samples of a physical object in 3D space. A PC video consists of temporal frames evenly spaced in time, each containing a static PC at one time instant. PCs in adjacent frames typically do not have point-to-point (P2P) correspondence, and thus exploiting temporal redundancy for PC restoration across frames is difficult. In this paper, we focus on the super-resolution (SR) problem for PC video: increase point density of PCs in video frames while preserving salient geometric features consistently across time. We accomplish this with two ideas. First, we establish partial P2P coupling between PCs of adjacent frames by interpolating interior points in a low-resolution PC patch in frame $t$ and translating them to a corresponding patch in frame $t+1$ , via a motion model computed by iterative closest point (ICP). Second, we promote piecewise smoothness in 3D geometry in each patch using feature graph Laplacian regularizer (FGLR) in an easily computable quadratic form. The two ideas translate to an unconstrained quadratic programming (QP) problem with a system of linear equations as solution—one where we ensure the numerical stability by upper-bounding the condition number of the coefficient matrix. Finally, to improve the accuracy of the ICP motion model, we re-sample points in a super-resolved patch at time $t$ to better match a low-resolution patch at time $t+1$ via bipartite graph matching after each SR iteration. Experimental results show temporally consistent super-resolved PC videos generated by our scheme, outperforming SR competitors that optimized on a per-frame basis, in two established PC metrics. [ABSTRACT FROM AUTHOR]

Most video-based person re-identification (re-id) methods only focus on appearance features but neglect motion features. In fact, motion features can help to distinguish the target persons that are hard to be identified only by appearance features. However, most existing temporal information modeling methods cannot extract motion features effectively or efficiently for v ideo-based re-id. In this paper, we propose a more efficient Motion Feature Aggregation (MFA) method to model and aggregate motion information in the feature map level for video-based re-id. The proposed MFA consists of (i) a coarse-grained motion learning module, which extracts coarse-grained motion features based on the position changes of body parts over time, and (ii) a fine-grained motion learning module, which extracts fine-grained motion features based on the appearance changes of body parts over time. These two modules can model motion information from different granularities and are complementary to each other. It is easy to combine the proposed method with existing network architectures for end-to-end training. Extensive experiments on four widely used datasets demonstrate that the motion features extracted by MFA are crucial complements to appearance features for video-based re-id, especially for the scenario with large appearance changes. Besides, the results on LS-VID, the current largest publicly available video-based re-id dataset, surpass the state-of-the-art methods by a large margin. The code is available at: https://github.com/guxinqian/Simple-ReID. [ABSTRACT FROM AUTHOR]

The success of current deep saliency models heavily depends on large amounts of annotated human fixation data to fit the highly non-linear mapping between the stimuli and visual saliency. Such fully supervised data-driven approaches are annotation-intensive and often fail to consider the underlying mechanisms of visual attention. In contrast, in this paper, we introduce a model based on various cognitive theories of visual saliency, which learns visual attention patterns in a weakly supervised manner. Our approach incorporates insights from cognitive science as differentiable submodules, resulting in a unified, end-to-end trainable framework. Specifically, our model encapsulates the following important components motivated from biological vision. (a) As scene semantics are closely related to visually attentive regions, our model encodes discriminative spatial information for scene understanding through spatial visual semantics embedding. (b) To model the objectness factors in visual attention deployment, we incorporate object-level semantics embedding and object relation information. (c) Considering the “winner-take-all” mechanism in visual stimuli processing, we model the competition mechanism among objects with softmax based neural attention. (d) Lastly, a conditional center prior is learned to mimic the spatial distribution bias of visual attention. Furthermore, we propose novel loss functions to utilize supervision cues from image-level semantics, saliency prior knowledge, and self-information compression. Experiments show that our method achieves promising results, and even outperforms many of its fully supervised counterparts. Overall, our weakly supervised saliency method makes an essential step towards reducing the annotation budget of current approaches, as well as providing a more comprehensive understanding of the visual attention mechanism. Our code is available at: https://github.com/ashleylqx/WeakFixation.git. [ABSTRACT FROM AUTHOR]

The Normalized Cut (NCut) model is a popular graph-based model for image segmentation. But it suffers from the excessive normalization problem and weakens the small object and twig segmentation. In this paper, we propose an Explored Normalized Cut (ENCut) model that establishes a balance graph model by adopting a meaningful-loop and a k-step random walk, which reduces the energy of small salient region, so as to enhance the small object segmentation. To improve the twig segmentation, our ENCut model is further enhanced by a new Random Walk Refining Term (RWRT) that adds local attention to our model with the help of an un-supervising random walk. Finally, a move-making based strategy is developed to efficiently solve the ENCut model with RWRT. Experiments on three standard datasets indicate that our model can achieve state-of-the-art results among the NCut-based segmentation models. [ABSTRACT FROM AUTHOR]

Tubular structure tracking is a crucial task in the fields of computer vision and medical image analysis. The minimal paths-based approaches have exhibited their strong ability in tracing tubular structures, by which a tubular structure can be naturally modeled as a minimal geodesic path computed with a suitable geodesic metric. However, existing minimal paths-based tracing approaches still suffer from difficulties such as the shortcuts and short branches combination problems, especially when dealing with the images involving complicated tubular tree structures or background. In this paper, we introduce a new minimal paths-based model for minimally interactive tubular structure centerline extraction in conjunction with a perceptual grouping scheme. Basically, we take into account the prescribed tubular trajectories and curvature-penalized geodesic paths to seek suitable shortest paths. The proposed approach can benefit from the local smoothness prior on tubular structures and the global optimality of the used graph-based path searching scheme. Experimental results on both synthetic and real images prove that the proposed model indeed obtains outperformance comparing with the state-of-the-art minimal paths-based tubular structure tracing algorithms. [ABSTRACT FROM AUTHOR]

Deconvolutional networks have attracted extensive attention and have been successfully applied in the field of computer vision. In this paper we propose a novel two-branch deconvolutional network (TBDN) that can improve the performance of conventional deconvolutional networks and reduce the computational complexity. A feasible iterative algorithm is designed to solve the optimization problem for the TBDN model, and a theoretical analysis of the convergence and computational complexity for the algorithm is also provided. The application of the TBDN in stereo matching is presented by constructing a disparity estimation network. Extensive experimental results on four commonly used datasets demonstrate the efficiency and effectiveness of the proposed TBDN. [ABSTRACT FROM AUTHOR]

Recent studies have pointed out that many well-developed Visual Question Answering (VQA) models are heavily affected by the language prior problem. It refers to making predictions based on the co-occurrence pattern between textual questions and answers instead of reasoning upon visual contents. To tackle this problem, most existing methods focus on strengthening the visual feature learning capability to reduce this text shortcut influence on model decisions. However, few efforts have been devoted to analyzing its inherent cause and providing an explicit interpretation. It thus lacks a good guidance for the research community to move forward in a purposeful way, resulting in model construction perplexity towards overcoming this non-trivial problem. In this paper, we propose to interpret the language prior problem in VQA from a class-imbalance view. Concretely, we design a novel interpretation scheme whereby the loss of mis-predicted frequent and sparse answers from the same question type is distinctly exhibited during the late training phase. It explicitly reveals why the VQA model tends to produce a frequent yet obviously wrong answer, to a given question whose right answer is sparse in the training set. Based upon this observation, we further propose a novel loss re-scaling approach to assign different weights to each answer according to the training data statistics for estimating the final loss. We apply our approach into six strong baselines and the experimental results on two VQA-CP benchmark datasets evidently demonstrate its effectiveness. In addition, we also justify the validity of the class imbalance interpretation scheme on other computer vision tasks, such as face recognition and image classification. [ABSTRACT FROM AUTHOR]

Contrastive learning has revolutionized the field of computer vision, learning rich representations from unlabeled data, which generalize well to diverse vision tasks. Consequently, it has become increasingly important to explain these approaches and understand their inner workings mechanisms. Given that contrastive models are trained with interdependent and interacting inputs and aim to learn invariance through data augmentation, the existing methods for explaining single-image systems (e.g., image classification models) are inadequate as they fail to account for these factors and typically assume independent inputs. Additionally, there is a lack of evaluation metrics designed to assess pairs of explanations, and no analytical studies have been conducted to investigate the effectiveness of different techniques used to explaining contrastive learning. In this work, we design visual explanation methods that contribute towards understanding similarity learning tasks from pairs of images. We further adapt existing metrics, used to evaluate visual explanations of image classification systems, to suit pairs of explanations and evaluate our proposed methods with these metrics. Finally, we present a thorough analysis of visual explainability methods for contrastive learning, establish their correlation with downstream tasks and demonstrate the potential of our approaches to investigate their merits and drawbacks. [ABSTRACT FROM AUTHOR]

The Visual Multimethod Assessment Fusion (VMAF) algorithm has recently emerged as a state-of-the-art approach to video quality prediction, that now pervades the streaming and social media industry. However, since VMAF requires the evaluation of a heterogeneous set of quality models, it is computationally expensive. Given other advances in hardware-accelerated encoding, quality assessment is emerging as a significant bottleneck in video compression pipelines. Towards alleviating this burden, we propose a novel Fusion of Unified Quality Evaluators (FUNQUE) framework, by enabling computation sharing and by using a transform that is sensitive to visual perception to boost accuracy. Further, we expand the FUNQUE framework to define a collection of improved low-complexity fused-feature models that advance the state-of-the-art of video quality performance with respect to both accuracy, by 4.2% to 5.3%, and computational efficiency, by factors of 3.8 to 11 times! [ABSTRACT FROM AUTHOR]

As an important yet challenging task in Earth observation, change detection (CD) is undergoing a technological revolution, given the broadening application of deep learning. Nevertheless, existing deep learning-based CD methods still suffer from two salient issues: 1) incomplete temporal modeling, and 2) space-time coupling. In view of these issues, we propose a more explicit and sophisticated modeling of time and accordingly establish a pair-to-video change detection (P2V-CD) framework. First, a pseudo transition video that carries rich temporal information is constructed from the input image pair, interpreting CD as a problem of video understanding. Then, two decoupled encoders are utilized to spatially and temporally recognize the type of transition, and the encoders are laterally connected for mutual promotion. Furthermore, the deep supervision technique is applied to accelerate the model training. We illustrate experimentally that the P2V-CD method compares favorably to other state-of-the-art CD approaches in terms of both the visual effect and the evaluation metrics, with a moderate model size and relatively lower computational overhead. Extensive feature map visualization experiments demonstrate how our method works beyond making contrasts between bi-temporal images. Source code is available at https://github.com/Bobholamovic/CDLab. [ABSTRACT FROM AUTHOR]

Most existing RGB-D salient object detection (SOD) models adopt a two-stream structure to extract the information from the input RGB and depth images. Since they use two subnetworks for unimodal feature extraction and multiple multi-modal feature fusion modules for extracting cross-modal complementary information, these models require a huge number of parameters, thus hindering their real-life applications. To remedy this situation, we propose a novel middle-level feature fusion structure that allows to design a lightweight RGB-D SOD model. Specifically, the proposed structure first employs two shallow subnetworks to extract low- and middle-level unimodal RGB and depth features, respectively. Afterward, instead of integrating middle-level unimodal features multiple times at different layers, we just fuse them once via a specially designed fusion module. On top of that, high-level multi-modal semantic features are further extracted for final salient object detection via an additional subnetwork. This will greatly reduce the network’s parameters. Moreover, to compensate for the performance loss due to parameter deduction, a relation-aware multi-modal feature fusion module is specially designed to effectively capture the cross-modal complementary information during the fusion of middle-level multi-modal features. By enabling the feature-level and decision-level information to interact, we maximize the usage of the fused cross-modal middle-level features and the extracted cross-modal high-level features for saliency prediction. Experimental results on several benchmark datasets verify the effectiveness and superiority of the proposed method over some state-of-the-art methods. Remarkably, our proposed model has only 3.9M parameters and runs at 33 FPS. [ABSTRACT FROM AUTHOR]

Vision-based fire detection systems have been significantly improved by deep models; however, higher numbers of false alarms and a slow inference speed still hinder their practical applicability in real-world scenarios. For a balanced trade-off between computational cost and accuracy, we introduce dual fire attention network (DFAN) to achieve effective yet efficient fire detection. The first attention mechanism highlights the most important channels from the features of an existing backbone model, yielding significantly emphasized feature maps. Then, a modified spatial attention mechanism is employed to capture spatial details and enhance the discrimination potential of fire and non-fire objects. We further optimize the DFAN for real-world applications by discarding a significant number of extra parameters using a meta-heuristic approach, which yields around 50% higher FPS values. Finally, we contribute a medium-scale challenging fire classification dataset by considering extremely diverse, highly similar fire/non-fire images and imbalanced classes, among many other complexities. The proposed dataset advances the traditional fire detection datasets by considering multiple classes to answer the following question: what is on fire? We perform experiments on four widely used fire detection datasets, and the DFAN provides the best results compared to 21 state-of-the-art methods. Consequently, our research provides a baseline for fire detection over edge devices with higher accuracy and better FPS values, and the proposed dataset extension provides indoor fire classes and a greater number of outdoor fire classes; these contributions can be used in significant future research. Our codes and dataset will be publicly available at https://github.com/tanveer-hussain/DFAN. [ABSTRACT FROM AUTHOR]

Although deep neural networks (DNNs) have been shown to be susceptible to image-agnostic adversarial attacks on natural image classification problems, the effects of such attacks on DNN-based texture recognition have yet to be explored. As part of our work, we find that limiting the perturbation’s $l_{p}$ norm in the spatial domain may not be a suitable way to restrict the perceptibility of universal adversarial perturbations for texture images. Based on the fact that human perception is affected by local visual frequency characteristics, we propose a frequency-tuned universal attack method to compute universal perturbations in the frequency domain. Our experiments indicate that our proposed method can produce less perceptible perturbations yet with a similar or higher white-box fooling rates on various DNN texture classifiers and texture datasets as compared to existing universal attack techniques. We also demonstrate that our approach can improve the attack robustness against defended models as well as the cross-dataset transferability for texture recognition problems. [ABSTRACT FROM AUTHOR]

Despite considerable progress, image captioning still suffers from the huge difference in quality between easy and hard examples, which is left unexploited in existing methods. To address this issue, we explore the hard example mining in image captioning, and propose a simple yet effective mechanism to instruct the model to pay more attention to hard examples, thereby improving the performance in both general and complex scenarios. We first propose a novel learning strategy, termed Metric-oriented Focal Mechanism (MFM), for hard example mining in image captioning. Differing from the existing strategies for classification tasks, MFM can adopt the generative metrics of image captioning to measure the difficulties of examples, and then up-weight the rewards of hard examples during training. To make MFM applicable to different datasets without tedious parameter tuning, we further introduce an adaptive reward metric called Effective CIDEr (ECIDEr), which considers the data distribution of easy and hard examples during reward estimation. Extensive experiments are conducted on the MS COCO benchmark, and the results show that while maintaining the performance on simple examples, MFM can significantly improve the quality of captions for hard examples. The ECIDEr-based MFM is equipped on the current SOTA method, e.g., DLCT (Luo et al., 2021), which outperforms all existing methods and achieves new state-of-the-art performance on both the off-line and on- line testing, i.e., 134.3 CIDEr for the off-line testing and 136.1 for the on- line testing of MSCOCO. To validate the generalization ability of ECIDEr-based MFM, we also apply it to another dataset, namely Flickr30k, and superior performance gains can also be obtained. [ABSTRACT FROM AUTHOR]

Recently, owing to the superior performances, knowledge distillation-based (kd-based) methods with the exemplar rehearsal have been widely applied in class incremental learning (CIL). However, we discover that they suffer from the feature uncalibration problem, which is caused by directly transferring knowledge from the old model immediately to the new model when learning a new task. As the old model confuses the feature representations between the learned and new classes, the kd loss and the classification loss used in kd-based methods are heterogeneous. This is detrimental if we learn the existing knowledge from the old model directly in the way as in typical kd-based methods. To tackle this problem, the feature calibration network (FCN) is proposed, which is used to calibrate the existing knowledge to alleviate the feature representation confusion of the old model. In addition, to relieve the task-recency bias of FCN caused by the limited storage memory in CIL, we propose a novel image-feature hybrid sample rehearsal strategy to train FCN by splitting the memory budget to store the image-and-feature exemplars of the previous tasks. As feature embeddings of images have much lower-dimensions, this allows us to store more samples to train FCN. Based on these two improvements, we propose the Cascaded Knowledge Distillation Framework (CKDF) including three main stages. The first stage is used to train FCN to calibrate the existing knowledge of the old model. Then, the new model is trained simultaneously by transferring knowledge from the calibrated teacher model through the knowledge distillation strategy and learning new classes. Finally, after completing the new task learning, the feature exemplars of previous tasks are updated. Importantly, we demonstrate that the proposed CKDF is a general framework that can be applied to various kd-based methods. Experimental results show that our method achieves state-of-the-art performances on several CIL benchmarks. [ABSTRACT FROM AUTHOR]

Image restoration is a long-standing problem in signal processing and low-level computer vision. Previous studies have shown that imposing a low-rank Tucker decomposition (TKD) constraint could produce impressive performances. However, the TKD-based schemes may lead to the overfitting/underfitting problem because of incorrectly predefined ranks. To address this issue, we prove that the $n$ -rank is upper bounded by the rank of each Tucker factor matrix. Using this relationship, we propose a formulation by imposing the nuclear norm regularization on the latent factors of TKD, which can avoid the burden of rank selection and reduce the computational cost when dealing with large-scale tensors. In this formulation, we adopt the Minimax Concave Penalty to remove the impulsive noise instead of the $l_{1}$ -norm which may deviate from both the data-acquisition model and the prior model. Moreover, we employ an anisotropic total variation regularization to explore the piecewise smooth structure in both spatial and spectral domains. To solve this problem, we design the symmetric Gauss-Seidel (sGS) based alternating direction method of multipliers (ADMM) algorithm. Compared to the directly extended ADMM, our algorithm can achieve higher accuracy since more structural information is utilized. Finally, we conduct experiments on the three kinds of datasets, numerical results demonstrate the superiority of the proposed method, especially, the average PSNR of the proposed method can improve about 1~5dB for each noise level of color images. [ABSTRACT FROM AUTHOR]

Despite the exciting performance, Transformer is criticized for its excessive parameters and computation cost. However, compressing Transformer remains as an open problem due to its internal complexity of the layer designs, i.e., Multi-Head Attention (MHA) and Feed-Forward Network (FFN). To address this issue, we introduce Group-wise Transformation towards a universal yet lightweight Transformer for vision-and-language tasks, termed as LW-Transformer. LW-Transformer applies Group-wise Transformation to reduce both the parameters and computations of Transformer, while also preserving its two main properties, i.e., the efficient attention modeling on diverse subspaces of MHA, and the expanding-scaling feature transformation of FFN. We apply LW-Transformer to a set of Transformer-based networks, and quantitatively measure them on three vision-and-language tasks and six benchmark datasets. Experimental results show that while saving a large number of parameters and computations, LW-Transformer achieves very competitive performance against the original Transformer networks for vision-and-language tasks. To examine the generalization ability, we apply LW-Transformer to the task of image classification, and build its network based on a recently proposed image Transformer called Swin-Transformer, where the effectiveness can be also confirmed. [ABSTRACT FROM AUTHOR]

Recently fast arbitrary-shaped text detection has become an attractive research topic. However, most existing methods are non-real-time, which may fall short in intelligent systems. Although a few real-time text methods are proposed, the detection accuracy is far behind non-real-time methods. To improve the detection accuracy and speed simultaneously, we propose a novel fast and accurate text detection framework, namely CM-Net, which is constructed based on a new text representation method and a multi-perspective feature (MPF) module. The former can fit arbitrary-shaped text contours by concentric mask (CM) in an efficient and robust way. The latter encourages the network to learn more CM-related discriminative features from multiple perspectives and brings no extra computational cost. Benefiting the advantages of CM and MPF, the proposed CM-Net only needs to predict one CM of the text instance to rebuild the text contour and achieves the best balance between detection accuracy and speed compared with previous works. Moreover, to ensure that multi-perspective features are effectively learned, the multi-factor constraints loss is proposed. Extensive experiments demonstrate the proposed CM is efficient and robust to fit arbitrary-shaped text instances, and also validate the effectiveness of MPF and constraints loss for discriminative text features recognition. Furthermore, experimental results show that the proposed CM-Net is superior to existing state-of-the-art (SOTA) real-time text detection methods in both detection speed and accuracy on MSRA-TD500, CTW1500, Total-Text, and ICDAR2015 datasets. [ABSTRACT FROM AUTHOR]

Indirect methods for visual SLAM are gaining popularity due to their robustness to environmental variations. ORB-SLAM2 (Mur-Artal and Tardós, 2017) is a benchmark method in this domain, however, it consumes significant time for computing descriptors that never get reused unless a frame is selected as a keyframe. To overcome these problems, we present FastORB-SLAM which is light-weight and efficient as it tracks keypoints between adjacent frames without computing descriptors. To achieve this, a two stage descriptor-independent keypoint matching method is proposed based on sparse optical flow. In the first stage, we predict initial keypoint correspondences via a simple but effective motion model and then robustly establish the correspondences via pyramid-based sparse optical flow tracking. In the second stage, we leverage the constraints of the motion smoothness and epipolar geometry to refine the correspondences. In particular, our method computes descriptors only for keyframes. We test FastORB-SLAM on TUM and ICL-NUIM RGB-D datasets and compare its accuracy and efficiency to nine existing RGB-D SLAM methods. Qualitative and quantitative results show that our method achieves state-of-the-art accuracy and is about twice as fast as the ORB-SLAM2. [ABSTRACT FROM AUTHOR]

Hashing has been widely applied to the large-scale approximate nearest neighbor search problem owing to its high efficiency and low storage requirement. Most investigations concentrate on learning hashing methods in a centralized setting. However, in existing big data systems, data is often stored across different nodes. In some situations, data is even collected in a distributed manner. A straightforward way to solve this problem is to aggregate all the data into the fusion center to obtain the search result (aggregating method). However, this strategy is not feasible because of the prohibitive communication cost. Although a few distributed hashing methods have been proposed to reduce this cost, they only focus on designing a distributed algorithm for a specific global optimization objective without considering scalability. Moreover, existing distributed hashing methods aim at finding a distributed solution to hashing, meanwhile avoiding accuracy loss, rather than improving accuracy. To address these challenges, we propose a Scalable Distributed Hashing (SDisH) model in which most existing hashing methods can be extended to process distributed data with no changes. Furthermore, to improve accuracy, we utilize the search radius as a global variable across different nodes to achieve a global optimum search result for every iteration. In addition, a voting algorithm is presented based on the results produced by multiple iterations to further reduce search errors. Theoretical analyses of communication, computation, and accuracy demonstrate the superiority of the proposed model. Numerical simulations on three large-scale and two relatively small benchmark datasets also show that the SDisH model achieves up to 44.75% and 10.23% accuracy gains compared to the aggregating method and state-of-the-art distributed hashing methods, respectively. [ABSTRACT FROM AUTHOR]

In this work, we address the challenging problem of completely blind video quality assessment (BVQA) of user generated content (UGC). The challenge is twofold since the quality prediction model is oblivious of human opinion scores, and there are no well-defined distortion models for UGC content. Our solution is inspired by a recent computational neuroscience model which hypothesizes that the human visual system (HVS) transforms a natural video input to follow a straighter temporal trajectory in the perceptual domain. A bandpass filter based computational model of the lateral geniculate nucleus (LGN) and V1 regions of the HVS was used to validate the perceptual straightening hypothesis. We hypothesize that distortions in natural videos lead to loss in straightness (or increased curvature) in their transformed representations in the HVS. We provide extensive empirical evidence to validate our hypothesis. We quantify the loss in straightness as a measure of temporal quality, and show that this measure delivers acceptable quality prediction performance on its own. Further, the temporal quality measure is combined with a state-of-the-art blind spatial (image) quality metric to design a blind video quality predictor that we call STraightness Evaluation Metric (STEM). STEM is shown to deliver state-of-the-art performance over the class of BVQA algorithms on five UGC VQA datasets including KoNViD-1K, LIVE-Qualcomm, LIVE-VQC, CVD and YouTube-UGC. Importantly, our solution is completely blind i.e., training-free, generalizes very well, is explainable, has few tunable parameters, and is simple and easy to implement. [ABSTRACT FROM AUTHOR]