Untrimmed videos on social media or those captured by robots and surveillance cameras are of varied aspect ratios. However, 3D CNNs usually require as input a square-shaped video, whose spatial dimension is smaller than the original. Random- or center-cropping may leave out the video's subject altogether. To address this, we propose an unsupervised video cropping approach by shaping this as a retargeting and video-to-video synthesis problem. The synthesized video maintains a 1:1 aspect ratio, is smaller in size and is targeted at video-subject(s) throughout the entire duration. First, action localization is performed on each frame by identifying patches with homogeneous motion patterns. Thus, a single salient patch is pinpointed per frame. But to avoid viewpoint jitters and flickering, any inter-frame scale or position changes among the patches should be performed gradually over time. This issue is addressed with a polyBezier fitting in 3D space that passes through some chosen pivot timestamps and whose shape is influenced by the in-between control timestamps. To corroborate the effectiveness of the proposed method, we evaluate the video classification task by comparing our dynamic cropping technique with random cropping on three benchmark datasets, viz. UCF-101, HMDB-51 and ActivityNet v1.3. The clip and top-1 accuracy for video classification after our cropping, outperform 3D CNN performances for same-sized random-crop inputs, also surpassing some larger random-crop sizes.
Part-level Action Parsing aims at part state parsing for boosting action recognition in videos. Despite of dramatic progresses in the area of video classification research, a severe problem faced by the community is that the detailed understanding of human actions is ignored. Our motivation is that parsing human actions needs to build models that focus on the specific problem. We present a simple yet effective approach, named disentangled action parsing (DAP). Specifically, we divided the part-level action parsing into three stages: 1) person detection, where a person detector is adopted to detect all persons from videos as well as performs instance-level action recognition; 2) Part parsing, where a part-parsing model is proposed to recognize human parts from detected person images; and 3) Action parsing, where a multi-modal action parsing network is used to parse action category conditioning on all detection results that are obtained from previous stages. With these three major models applied, our approach of DAP records a global mean of $0.605$ score in 2021 Kinetics-TPS Challenge.
Pooling layers are essential building blocks of convolutional neural networks (CNNs), to reduce computational overhead and increase the receptive fields of proceeding convolutional operations. Their goal is to produce downsampled volumes that closely resemble the input volume while, ideally, also being computationally and memory efficient. Meeting both these requirements remains a challenge. To this end, we propose an adaptive and exponentially weighted pooling method: adaPool. Our method learns a regional-specific fusion of two sets of pooling kernels that are based on the exponent of the Dice-Sorensen coefficient and the exponential maximum, respectively. AdaPool improves the preservation of detail on a range of tasks including image and video classification and object detection. A key property of adaPool is its bidirectional nature. In contrast to common pooling methods, the learned weights can also be used to upsample activation maps. We term this method adaUnPool. We evaluate adaUnPool on image and video super-resolution and frame interpolation. For benchmarking, we introduce Inter4K, a novel high-quality, high frame-rate video dataset. Our experiments demonstrate that adaPool systematically achieves better results across tasks and backbones, while introducing a minor additional computational and memory overhead.
The existing few-shot video classification methods often employ a meta-learning paradigm by designing customized temporal alignment module for similarity calculation. While significant progress has been made, these methods fail to focus on learning effective representations, and heavily rely on the ImageNet pre-training, which might be unreasonable for the few-shot recognition setting due to semantics overlap. In this paper, we aim to present an in-depth study on few-shot video classification by making three contributions. First, we perform a consistent comparative study on the existing metric-based methods to figure out their limitations in representation learning. Accordingly, we propose a simple classifier-based baseline without any temporal alignment that surprisingly outperforms the state-of-the-art meta-learning based methods. Second, we discover that there is a high correlation between the novel action class and the ImageNet object class, which is problematic in the few-shot recognition setting. Our results show that the performance of training from scratch drops significantly, which implies that the existing benchmarks cannot provide enough base data. Finally, we present a new benchmark with more base data to facilitate future few-shot video classification without pre-training. The code will be made available at https://github.com/MCG-NJU/FSL-Video.
When compared to the image classification models, black-box adversarial attacks against video classification models have been largely understudied. This could be possible because, with video, the temporal dimension poses significant additional challenges in gradient estimation. Query-efficient black-box attacks rely on effectively estimated gradients towards maximizing the probability of misclassifying the target video. In this work, we demonstrate that such effective gradients can be searched for by parameterizing the temporal structure of the search space with geometric transformations. Specifically, we design a novel iterative algorithm Geometric TRAnsformed Perturbations (GEO-TRAP), for attacking video classification models. GEO-TRAP employs standard geometric transformation operations to reduce the search space for effective gradients into searching for a small group of parameters that define these operations. This group of parameters describes the geometric progression of gradients, resulting in a reduced and structured search space. Our algorithm inherently leads to successful perturbations with surprisingly few queries. For example, adversarial examples generated from GEO-TRAP have better attack success rates with ~73.55% fewer queries compared to the state-of-the-art method for video adversarial attacks on the widely used Jester dataset. Overall, our algorithm exposes vulnerabilities of diverse video classification models and achieves new state-of-the-art results under black-box settings on two large datasets. Code is available here: https://github.com/sli057/Geo-TRAP
Several studies have shown the relevance of biosignals in driver stress recognition. In this work, we examine something important that has been less frequently explored: We develop methods to test if the visual driving scene can be used to estimate a drivers' subjective stress levels. For this purpose, we use the AffectiveROAD video recordings and their corresponding stress labels, a continuous human-driver-provided stress metric. We use the common class discretization for stress, dividing its continuous values into three classes: low, medium, and high. We design and evaluate three computer vision modeling approaches to classify the driver's stress levels: (1) object presence features, where features are computed using automatic scene segmentation; (2) end-to-end image classification; and (3) end-to-end video classification. All three approaches show promising results, suggesting that it is possible to approximate the drivers' subjective stress from the information found in the visual scene. We observe that the video classification, which processes the temporal information integrated with the visual information, obtains the highest accuracy of $0.72$, compared to a random baseline accuracy of $0.33$ when tested on a set of nine drivers.
Multi-modal Ads Video Understanding Challenge is the first grand challenge aiming to comprehensively understand ads videos. Our challenge includes two tasks: video structuring in the temporal dimension and multi-modal video classification. It asks the participants to accurately predict both the scene boundaries and the multi-label categories of each scene based on a fine-grained and ads-related category hierarchy. Therefore, our task has four distinguishing features from previous ones: ads domain, multi-modal information, temporal segmentation, and multi-label classification. It will advance the foundation of ads video understanding and have a significant impact on many ads applications like video recommendation. This paper presents an overview of our challenge, including the background of ads videos, an elaborate description of task and dataset, evaluation protocol, and our proposed baseline. By ablating the key components of our baseline, we would like to reveal the main challenges of this task and provide useful guidance for future research of this area. In this paper, we give an extended version of our challenge overview. The dataset will be publicly available at https://algo.qq.com/.
A big part of achieving Artificial General Intelligence(AGI) is to build a machine that can see and listen like humans. Much work has focused on designing models for image classification, video classification, object detection, pose estimation, speech recognition, etc., and has achieved significant progress in recent years thanks to deep learning. However, understanding the world is not enough. An AI agent also needs to know how to talk, especially how to communicate with a human. While perception (vision, for example) is more common across animal species, the use of complicated language is unique to humans and is one of the most important aspects of intelligence. In this thesis, we focus on generating textual output given visual input. In Chapter 3, we focus on generating the referring expression, a text description for an object in the image so that a receiver can infer which object is being described. We use a comprehension machine to directly guide the generated referring expressions to be more discriminative. In Chapter 4, we introduce a method that encourages discriminability in image caption generation. We show that more discriminative captioning models generate more descriptive captions. In Chapter 5, we study how training objectives and sampling methods affect the models' ability to generate diverse captions. We find that a popular captioning training strategy will be detrimental to the diversity of generated captions. In Chapter 6, we propose a model that can control the length of generated captions. By changing the desired length, one can influence the style and descriptiveness of the captions. Finally, in Chapter 7, we rank/generate informative image tags according to their information utility. The proposed method better matches what humans think are the most important tags for the images.
The automatic detection of events in complex sports games like soccer and handball using positional or video data is of large interest in research and industry. One requirement is a fundamental understanding of underlying concepts, i.e., events that occur on the pitch. Previous work often deals only with so-called low-level events based on well-defined rules such as free kicks, free throws, or goals. High-level events, such as passes, are less frequently approached due to a lack of consistent definitions. This introduces a level of ambiguity that necessities careful validation when regarding event annotations. Yet, this validation step is usually neglected as the majority of studies adopt annotations from commercial providers on private datasets of unknown quality and focuses on soccer only. To address these issues, we present (1) a universal taxonomy that covers a wide range of low and high-level events for invasion games and is exemplarily refined to soccer and handball, and (2) release two multi-modal datasets comprising video and positional data with gold-standard annotations to foster research in fine-grained and ball-centered event spotting. Experiments on human performance demonstrate the robustness of the proposed taxonomy, and that disagreements and ambiguities in the annotation increase with the complexity of the event. An I3D model for video classification is adopted for event spotting and reveals the potential for benchmarking. Datasets are available at: https://github.com/mm4spa/eigd
In this work, an extensive review of literature in the field of gesture recognition carried out along with the implementation of a simple classification system for hand hygiene stages based on deep learning solutions. A subset of robust dataset that consist of handwashing gestures with two hands as well as one-hand gestures such as linear hand movement utilized. A pretrained neural network model, RES Net 50, with image net weights used for the classification of 3 categories: Linear hand movement, rub hands palm to palm and rub hands with fingers interlaced movement. Correct predictions made for the first two classes with > 60% accuracy. A complete dataset along with increased number of classes and training steps will be explored as a future work.
In this work, deep learning models are applied to a segment of a robust hand-washing dataset that has been created with the help of 30 volunteers. This work demonstrates the classification of presence of one hand, two hands and no hand in the scene based on transfer learning. The pre-trained model; simplest NN from Keras library is utilized to train the network with 704 images of hand gestures and the predictions are carried out for the input image. Due to the controlled and restricted dataset, 100% accuracy is achieved during the training with correct predictions for the input image. Complete handwashing dataset with dense models such as AlexNet for video classification for hand hygiene stages will be used in the future work.
Face anti-spoofing is an important task to protect the security of face recognition. Most of previous work either struggle to capture discriminative and generalizable feature or rely on auxiliary information which is unavailable for most of industrial product. Inspired by the video classification work, we propose an efficient two-stream model to capture the key differences between live and spoof faces, which takes multi-frames and RGB difference as input respectively. Feature pyramid modules with two opposite fusion directions and pyramid pooling modules are applied to enhance feature representation. We evaluate the proposed method on the datasets of Siw, Oulu-NPU, CASIA-MFSD and Replay-Attack. The results show that our model achieves the state-of-the-art results on most of datasets' protocol with much less parameter size.
Transformer achieves remarkable successes in understanding 1 and 2-dimensional signals (e.g., NLP and Image Content Understanding). As a potential alternative to convolutional neural networks, it shares merits of strong interpretability, high discriminative power on hyper-scale data, and flexibility in processing varying length inputs. However, its encoders naturally contain computational intensive operations such as pair-wise self-attention, incurring heavy computational burden when being applied on the complex 3-dimensional video signals. This paper presents Token Shift Module (i.e., TokShift), a novel, zero-parameter, zero-FLOPs operator, for modeling temporal relations within each transformer encoder. Specifically, the TokShift barely temporally shifts partial [Class] token features back-and-forth across adjacent frames. Then, we densely plug the module into each encoder of a plain 2D vision transformer for learning 3D video representation. It is worth noticing that our TokShift transformer is a pure convolutional-free video transformer pilot with computational efficiency for video understanding. Experiments on standard benchmarks verify its robustness, effectiveness, and efficiency. Particularly, with input clips of 8/12 frames, the TokShift transformer achieves SOTA precision: 79.83%/80.40% on the Kinetics-400, 66.56% on EGTEA-Gaze+, and 96.80% on UCF-101 datasets, comparable or better than existing SOTA convolutional counterparts. Our code is open-sourced in: https://github.com/VideoNetworks/TokShift-Transformer.
The goal of few-shot video classification is to learn a classification model with good generalization ability when trained with only a few labeled videos. However, it is difficult to learn discriminative feature representations for videos in such a setting. In this paper, we propose Temporal Alignment Prediction (TAP) based on sequence similarity learning for few-shot video classification. In order to obtain the similarity of a pair of videos, we predict the alignment scores between all pairs of temporal positions in the two videos with the temporal alignment prediction function. Besides, the inputs to this function are also equipped with the context information in the temporal domain. We evaluate TAP on two video classification benchmarks including Kinetics and Something-Something V2. The experimental results verify the effectiveness of TAP and show its superiority over state-of-the-art methods.
Detection of a violence event in surveillance systems is playing a significant role in law enforcement and city safety. The effectiveness of violence event detectors measures by the speed of response and the accuracy and the generality over different kind of video sources with a different format. Several studies worked on the violence detection with focus either on speed or accuracy or both but not taking into account the generality over different kind of video sources. In this paper, we proposed a real-time violence detector based on deep-learning methods. The proposed model consists of CNN as a spatial feature extractor and LSTM as temporal relation learning method with a focus on the three-factor (overall generality - accuracy - fast response time). The suggested model achieved 98% accuracy with speed of 131 frames/sec. Comparison of the accuracy and the speed of the proposed model with previous works illustrated that the proposed model provides the highest accuracy and the fastest speed among all the previous works in the field of violence detection.
Data augmentation is a commonly used approach to improving the generalization of deep learning models. Recent works show that learned data augmentation policies can achieve better generalization than hand-crafted ones. However, most of these works use unified augmentation policies for all samples in a dataset, which is observed not necessarily beneficial for all labels in multi-label classification tasks, i.e., some policies may have negative impacts on some labels while benefitting the others. To tackle this problem, we propose a novel Label-Based AutoAugmentation (LB-Aug) method for multi-label scenarios, where augmentation policies are generated with respect to labels by an augmentation-policy network. The policies are learned via reinforcement learning using policy gradient methods, providing a mapping from instance labels to their optimal augmentation policies. Numerical experiments show that our LB-Aug outperforms previous state-of-the-art augmentation methods by large margins in multiple benchmarks on image and video classification.
Domain adaptation (DA) approaches address domain shift and enable networks to be applied to different scenarios. Although various image DA approaches have been proposed in recent years, there is limited research towards video DA. This is partly due to the complexity in adapting the different modalities of features in videos, which includes the correlation features extracted as long-term dependencies of pixels across spatiotemporal dimensions. The correlation features are highly associated with action classes and proven their effectiveness in accurate video feature extraction through the supervised action recognition task. Yet correlation features of the same action would differ across domains due to domain shift. Therefore we propose a novel Adversarial Correlation Adaptation Network (ACAN) to align action videos by aligning pixel correlations. ACAN aims to minimize the distribution of correlation information, termed as Pixel Correlation Discrepancy (PCD). Additionally, video DA research is also limited by the lack of cross-domain video datasets with larger domain shifts. We, therefore, introduce a novel HMDB-ARID dataset with a larger domain shift caused by a larger statistical difference between domains. This dataset is built in an effort to leverage current datasets for dark video classification. Empirical results demonstrate the state-of-the-art performance of our proposed ACAN for both existing and the new video DA datasets.
Humans perceive the world by concurrently processing and fusing high-dimensional inputs from multiple modalities such as vision and audio. Machine perception models, in stark contrast, are typically modality-specific and optimised for unimodal benchmarks, and hence late-stage fusion of final representations or predictions from each modality (`late-fusion') is still a dominant paradigm for multimodal video classification. Instead, we introduce a novel transformer based architecture that uses `fusion bottlenecks' for modality fusion at multiple layers. Compared to traditional pairwise self-attention, our model forces information between different modalities to pass through a small number of bottleneck latents, requiring the model to collate and condense the most relevant information in each modality and only share what is necessary. We find that such a strategy improves fusion performance, at the same time reducing computational cost. We conduct thorough ablation studies, and achieve state-of-the-art results on multiple audio-visual classification benchmarks including Audioset, Epic-Kitchens and VGGSound. All code and models will be released.
With the rapid development of social media, tremendous videos with new classes are generated daily, which raise an urgent demand for video classification methods that can continuously update new classes while maintaining the knowledge of old videos with limited storage and computing resources. In this paper, we summarize this task as Class-Incremental Video Classification (CIVC) and propose a novel framework to address it. As a subarea of incremental learning tasks, the challenge of catastrophic forgetting is unavoidable in CIVC. To better alleviate it, we utilize some characteristics of videos. First, we decompose the spatio-temporal knowledge before distillation rather than treating it as a whole in the knowledge transfer process; trajectory is also used to refine the decomposition. Second, we propose a dual granularity exemplar selection method to select and store representative video instances of old classes and key-frames inside videos under a tight storage budget. We benchmark our method and previous SOTA class-incremental learning methods on Something-Something V2 and Kinetics datasets, and our method outperforms previous methods significantly.
The vision community is witnessing a modeling shift from CNNs to Transformers, where pure Transformer architectures have attained top accuracy on the major video recognition benchmarks. These video models are all built on Transformer layers that globally connect patches across the spatial and temporal dimensions. In this paper, we instead advocate an inductive bias of locality in video Transformers, which leads to a better speed-accuracy trade-off compared to previous approaches which compute self-attention globally even with spatial-temporal factorization. The locality of the proposed video architecture is realized by adapting the Swin Transformer designed for the image domain, while continuing to leverage the power of pre-trained image models. Our approach achieves state-of-the-art accuracy on a broad range of video recognition benchmarks, including on action recognition (84.9 top-1 accuracy on Kinetics-400 and 86.1 top-1 accuracy on Kinetics-600 with ~20x less pre-training data and ~3x smaller model size) and temporal modeling (69.6 top-1 accuracy on Something-Something v2). The code and models will be made publicly available at https://github.com/SwinTransformer/Video-Swin-Transformer.