Many interesting events in the real world are rare making preannotated machine learning ready videos a rarity in consequence. Thus, temporal activity detection models that are able to learn from a few examples are desirable. In this paper, we present a conceptually simple and general yet novel framework for few-shot temporal activity detection based on proposal regression which detects the start and end time of the activities in untrimmed videos. Our model is end-to-end trainable, takes into account the frame rate differences between few-shot activities and untrimmed test videos, and can benefit from additional few-shot examples. We experiment on three large scale benchmarks for temporal activity detection (ActivityNet1.2, ActivityNet1.3 and THUMOS14 datasets) in a few-shot setting. We also study the effect on performance of different amount of overlap with activities used to pretrain the video classification backbone and propose corrective measures for future works in this domain. Our code will be made available.
Spiking neural networks (SNNs) can be used in low-power and embedded systems (such as emerging neuromorphic chips) due to their event-based nature. Also, they have the advantage of low computation cost in contrast to conventional artificial neural networks (ANNs), while preserving ANN's properties. However, temporal coding in layers of convolutional spiking neural networks and other types of SNNs has yet to be studied. In this paper, we provide insight into spatio-temporal feature extraction of convolutional SNNs in experiments designed to exploit this property. The shallow convolutional SNN outperforms state-of-the-art spatio-temporal feature extractor methods such as C3D, ConvLstm, and similar networks. Furthermore, we present a new deep spiking architecture to tackle real-world problems (in particular classification tasks) which achieved superior performance compared to other SNN methods on NMNIST (99.6%), DVS-CIFAR10 (69.2%) and DVS-Gesture (96.7%) and ANN methods on UCF-101 (42.1%) and HMDB-51 (21.5%) datasets. It is also worth noting that the training process is implemented based on variation of spatio-temporal backpropagation explained in the paper.
Deep neural networks are known to be susceptible to adversarial noise, which are tiny and imperceptible perturbations. Most of previous work on adversarial attack mainly focus on image models, while the vulnerability of video models is less explored. In this paper, we aim to attack video models by utilizing intrinsic movement pattern and regional relative motion among video frames. We propose an effective motion-excited sampler to obtain motion-aware noise prior, which we term as sparked prior. Our sparked prior underlines frame correlations and utilizes video dynamics via relative motion. By using the sparked prior in gradient estimation, we can successfully attack a variety of video classification models with fewer number of queries. Extensive experimental results on four benchmark datasets validate the efficacy of our proposed method.
In this paper we challenge the common assumption that convolutional layers in modern CNNs are translation invariant. We show that CNNs can and will exploit the absolute spatial location by learning filters that respond exclusively to particular absolute locations by exploiting image boundary effects. Because modern CNNs filters have a huge receptive field, these boundary effects operate even far from the image boundary, allowing the network to exploit absolute spatial location all over the image. We give a simple solution to remove spatial location encoding which improves translation invariance and thus gives a stronger visual inductive bias which particularly benefits small data sets. We broadly demonstrate these benefits on several architectures and various applications such as image classification, patch matching, and two video classification datasets.
Trained on large datasets, deep learning (DL) can accurately classify videos into hundreds of diverse classes. However, video data is expensive to annotate. Zero-shot learning (ZSL) proposes one solution to this problem. ZSL trains a model once, and generalizes to new tasks whose classes are not present in the training dataset. We propose the first end-to-end algorithm for ZSL in video classification. Our training procedure builds on insights from recent video classification literature and uses a trainable 3D CNN to learn the visual features. This is in contrast to previous video ZSL methods, which use pretrained feature extractors. We also extend the current benchmarking paradigm: Previous techniques aim to make the test task unknown at training time but fall short of this goal. We encourage domain shift across training and test data and disallow tailoring a ZSL model to a specific test dataset. We outperform the state-of-the-art by a wide margin. Our code, evaluation procedure and model weights are available at github.com/bbrattoli/ZeroShotVideoClassification.
We propose a semi-supervised learning approach for video classification, VideoSSL, using convolutional neural networks (CNN). Like other computer vision tasks, existing supervised video classification methods demand a large amount of labeled data to attain good performance. However, annotation of a large dataset is expensive and time consuming. To minimize the dependence on a large annotated dataset, our proposed semi-supervised method trains from a small number of labeled examples and exploits two regulatory signals from unlabeled data. The first signal is the pseudo-labels of unlabeled examples computed from the confidences of the CNN being trained. The other is the normalized probabilities, as predicted by an image classifier CNN, that captures the information about appearances of the interesting objects in the video. We show that, under the supervision of these guiding signals from unlabeled examples, a video classification CNN can achieve impressive performances utilizing a small fraction of annotated examples on three publicly available datasets: UCF101, HMDB51 and Kinetics.
In this paper, we propose a new video representation learning method, named Temporal Squeeze (TS) pooling, which can extract the essential movement information from a long sequence of video frames and map it into a set of few images, named Squeezed Images. By embedding the Temporal Squeeze pooling as a layer into off-the-shelf Convolution Neural Networks (CNN), we design a new video classification model, named Temporal Squeeze Network (TeSNet). The resulting Squeezed Images contain the essential movement information from the video frames, corresponding to the optimization of the video classification task. We evaluate our architecture on two video classification benchmarks, and the results achieved are compared to the state-of-the-art.
Deep neural networks for video classification, just like image classification networks, may be subjected to adversarial manipulation. The main difference between image classifiers and video classifiers is that the latter usually use temporal information contained within the video. In this work we present a manipulation scheme for fooling video classifiers by introducing a flickering temporal perturbation that in some cases may be unnoticeable by human observers and is implementable in the real world. After demonstrating the manipulation of action classification of single videos, we generalize the procedure to make universal adversarial perturbation, achieving high fooling ratio. In addition, we generalize the universal perturbation and produce a temporal-invariant perturbation, which can be applied to the video without synchronizing the perturbation to the input. The attack was implemented on several target models and the transferability of the attack was demonstrated. These properties allow us to bridge the gap between simulated environment and real-world application, as will be demonstrated in this paper for the first time for an over-the-air flickering attack.
Action recognition is an important and challenging problem in video analysis. Although the past decade has witnessed progress in action recognition with the development of deep learning, such process has been slow in competitive sports content analysis. To promote the research on action recognition from competitive sports video clips, we introduce a Figure Skating Dataset (FSD-10) for finegrained sports content analysis. To this end, we collect 1484 clips from the worldwide figure skating championships in 2017-2018, which consist of 10 different actions in men/ladies programs. Each clip is at a rate of 30 frames per second with resolution 1080 $\times$ 720. These clips are then annotated by experts in type, grade of execution, skater info, .etc. To build a baseline for action recognition in figure skating, we evaluate state-of-the-art action recognition methods on FSD-10. Motivated by the idea that domain knowledge is of great concern in sports field, we propose a keyframe based temporal segment network (KTSN) for classification and achieve remarkable performance. Experimental results demonstrate that FSD-10 is an ideal dataset for benchmarking action recognition algorithms, as it requires to accurately extract action motions rather than action poses. We hope FSD-10, which is designed to have a large collection of finegrained actions, can serve as a new challenge to develop more robust and advanced action recognition models.
In this report, the method for the iqiyi submission to the task of ActivityNet 2019 Kinetics-700 challenge is described. Three models are involved in the model ensemble stage: TSN, HG-NL and StNet. We propose the hierarchical group-wise non-local (HG-NL) module for frame-level features aggregation for video classification. The standard non-local (NL) module is effective in aggregating frame-level features on the task of video classification but presents low parameters efficiency and high computational cost. The HG-NL method involves a hierarchical group-wise structure and generates multiple attention maps to enhance performance. Basing on this hierarchical group-wise structure, the proposed method has competitive accuracy, fewer parameters and smaller computational cost than the standard NL. For the task of ActivityNet 2019 Kinetics-700 challenge, after model ensemble, we finally obtain an averaged top-1 and top-5 error percentage 28.444% on the test set.
Multi-label image and video classification are fundamental yet challenging tasks in computer vision. The main challenges lie in capturing spatial or temporal dependencies between labels and discovering the locations of discriminative features for each class. In order to overcome these challenges, we propose to use cross-modality attention with semantic graph embedding for multi label classification. Based on the constructed label graph, we propose an adjacency-based similarity graph embedding method to learn semantic label embeddings, which explicitly exploit label relationships. Then our novel cross-modality attention maps are generated with the guidance of learned label embeddings. Experiments on two multi-label image classification datasets (MS-COCO and NUS-WIDE) show our method outperforms other existing state-of-the-arts. In addition, we validate our method on a large multi-label video classification dataset (YouTube-8M Segments) and the evaluation results demonstrate the generalization capability of our method.
There have been many efforts in attacking image classification models with adversarial perturbations, but the same topic on video classification has not yet been thoroughly studied. This paper presents a novel idea of video-based attack, which appends a few dummy frames (e.g., containing the texts of `thanks for watching') to a video clip and then adds adversarial perturbations only on these new frames. Our approach enjoys three major benefits, namely, a high success rate, a low perceptibility, and a strong ability in transferring across different networks. These benefits mostly come from the common dummy frame which pushes all samples towards the boundary of classification. On the other hand, such attacks are easily to be concealed since most people would not notice the abnormality behind the perturbed video clips. We perform experiments on two popular datasets with six state-of-the-art video classification models, and demonstrate the effectiveness of our approach in the scenario of universal video attacks.
Action Detection is a complex task that aims to detect and classify human actions in video clips. Typically, it has been addressed by processing fine-grained features extracted from a video classification backbone. Recently, thanks to the robustness of object and people detectors, a deeper focus has been added on relationship modelling. Following this line, we propose a graph-based framework to learn high-level interactions between people and objects, in both space and time. In our formulation, spatio-temporal relationships are learned through self-attention on a multi-layer graph structure which can connect entities from consecutive clips, thus considering long-range spatial and temporal dependencies. The proposed module is backbone independent by design and does not require end-to-end training. Extensive experiments are conducted on the AVA dataset, where our model demonstrates state-of-the-art results and consistent improvements over baselines built with different backbones. Code is publicly available at https://github.com/aimagelab/STAGE_action_detection.
This paper introduces video domain generalization where most video classification networks degenerate due to the lack of exposure to the target domains of divergent distributions. We observe that the global temporal features are less generalizable, due to the temporal domain shift that videos from other unseen domains may have an unexpected absence or misalignment of the temporal relations. This finding has motivated us to solve video domain generalization by effectively learning the local-relation features of different timescales that are more generalizable, and exploiting them along with the global-relation features to maintain the discriminability. This paper presents the VideoDG framework with two technical contributions. The first is a new deep architecture named the Adversarial Pyramid Network, which improves the generalizability of video features by capturing the local-relation, global-relation, and cross-relation features progressively. On the basis of pyramid features, the second contribution is a new and robust approach of adversarial data augmentation that can bridge different video domains by improving the diversity and quality of augmented data. We construct three video domain generalization benchmarks in which domains are divided according to different datasets, different consequences of actions, or different camera views, respectively. VideoDG consistently outperforms the combinations of previous video classification models and existing domain generalization methods on all benchmarks.
There are many realistic applications of activity recognition where the set of potential activity descriptions is combinatorially large. This makes end-to-end supervised training of a recognition system impractical as no training set is practically able to encompass the entire label set. In this paper, we present an approach to fine-grained recognition that models activities as compositions of dynamic action signatures. This compositional approach allows us to reframe fine-grained recognition as zero-shot activity recognition, where a detector is composed "on the fly" from simple first-principles state machines supported by deep-learned components. We evaluate our method on the Olympic Sports and UCF101 datasets, where our model establishes a new state of the art under multiple experimental paradigms. We also extend this method to form a unique framework for zero-shot joint segmentation and classification of activities in video and demonstrate the first results in zero-shot decoding of complex action sequences on a widely-used surgical dataset. Lastly, we show that we can use off-the-shelf object detectors to recognize activities in completely de-novo settings with no additional training.
Training competitive deep video models is an order of magnitude slower than training their counterpart image models. Slow training causes long research cycles, which hinders progress in video understanding research. Following standard practice for training image models, video model training assumes a fixed mini-batch shape: a specific number of clips, frames, and spatial size. However, what is the optimal shape? High resolution models perform well, but train slowly. Low resolution models train faster, but they are inaccurate. Inspired by multigrid methods in numerical optimization, we propose to use variable mini-batch shapes with different spatial-temporal resolutions that are varied according to a schedule. The different shapes arise from resampling the training data on multiple sampling grids. Training is accelerated by scaling up the mini-batch size and learning rate when shrinking the other dimensions. We empirically demonstrate a general and robust grid schedule that yields a significant out-of-the-box training speedup without a loss in accuracy for different models (I3D, non-local, SlowFast), datasets (Kinetics, Something-Something, Charades), and training settings (with and without pre-training, 128 GPUs or 1 GPU). As an illustrative example, the proposed multigrid method trains a ResNet-50 SlowFast network 4.5x faster (wall-clock time, same hardware) while also improving accuracy (+0.8% absolute) on Kinetics-400 compared to the baseline training method. Code is available online.
The nonlocal-based blocks are designed for capturing long-range spatial-temporal dependencies in computer vision tasks. Although having shown excellent performances, they lack the mechanism to encode the rich, structured information among elements in an image. In this paper, to theoretically analyze the property of these nonlocal-based blocks, we provide a unified approach to interpreting them, where we view them as a graph filter generated on a fully-connected graph. When the graph filter is approximated by Chebyshev polynomials, a generalized formulation can be derived for explaining the existing nonlocal-based blocks ($\mathit{e.g.,}$ nonlocal block, nonlocal stage, double attention block). Furthermore, we propose an efficient and robust spectral nonlocal block, which can be flexibly inserted into deep neural networks to catch the long-range dependencies between spatial pixels or temporal frames. Experimental results demonstrate the clear-cut improvements and practical applicabilities of the spectral nonlocal block on image classification (Cifar-10/100, ImageNet), fine-grained image classification (CUB-200), action recognition (UCF-101), and person re-identification (ILID-SVID, Mars, Prid-2011) tasks.
Recently, Weakly-supervised Temporal Action Localization (WTAL) has been densely studied but there is still a large gap between weakly-supervised models and fully-supervised models. It is practical and intuitive to annotate temporal boundaries of a few examples and utilize them to help WTAL models better detect actions. However, the train-test discrepancy of action localization strategy prevents WTAL models from leveraging semi-supervision for further improvement. At training time, attention or multiple instance learning is used to aggregate predictions of each snippet for video-level classification; at test time, they first obtain action score sequences over time, then truncate segments of scores higher than a fixed threshold, and post-process action segments. The inconsistent strategy makes it hard to explicitly supervise the action localization model with temporal boundary annotations at training time. In this paper, we propose a Train-Test Consistent framework, TTC-Loc. In both training and testing time, our TTC-Loc localizes actions by comparing scores of action classes and predicted threshold, which enables it to be trained with semi-supervision. By fixing the train-test discrepancy, our TTC-Loc significantly outperforms the state-of-the-art performance on THUMOS'14, ActivityNet 1.2 and 1.3 when only video-level labels are provided for training. With full annotations of only one video per class and video-level labels for the other videos, our TTC-Loc further boosts the performance and achieves 33.4\% mAP (IoU threshold 0.5) on THUMOS's 14.
There is an urgent need for an effective video classification method by means of a small number of samples. The deficiency of samples could be effectively alleviated by generating samples through Generative Adversarial Networks (GAN), but the generation of videos on a typical category remains to be underexplored since the complex actions and the changeable viewpoints are difficult to simulate. In this paper, we propose a generative data augmentation method for temporal stream of the Temporal Segment Networks with the dynamic image. The dynamic image compresses the motion information of video into a still image, removing the interference factors such as the background. Thus it is easier to generate images with categorical motion information using GAN. We use the generated dynamic images to enhance the features, with regularization achieved as well, thereby to achieve the effect of video augmentation. In order to deal with the uneven quality of generated images, we propose a Self-Paced Selection (SPS) method, which automatically selects the high-quality generated samples to be added to the network training. Our method is verified on two benchmark datasets, HMDB51 and UCF101. The experimental results show that the method can improve the accuracy of video classification under the circumstance of sample insufficiency and sample imbalance.
In this work, we propose a generally applicable transformation unit for visual recognition with deep convolutional neural networks. This transformation explicitly models channel relationships with explainable control variables. These variables determine the neuron behaviors of competition or cooperation, and they are jointly optimized with the convolutional weight towards more accurate recognition. In Squeeze-and-Excitation (SE) Networks, the channel relationships are implicitly learned by fully connected layers, and the SE block is integrated at the block-level. We instead introduce a channel normalization layer to reduce the number of parameters and computational complexity. This lightweight layer incorporates a simple l2 normalization, enabling our transformation unit applicable to operator-level without much increase of additional parameters. Extensive experiments demonstrate the effectiveness of our unit with clear margins on many vision tasks, i.e., image classification on ImageNet, object detection and instance segmentation on COCO, video classification on Kinetics.