Adversarial robustness of deep neural networks has been actively investigated. However, most existing defense approaches are limited to a specific type of adversarial perturbations. Specifically, they often fail to offer resistance to multiple attack types simultaneously, i.e., they lack multi-perturbation robustness. Furthermore, compared to image recognition problems, the adversarial robustness of video recognition models is relatively unexplored. While several studies have proposed how to generate adversarial videos, only a handful of approaches about defense strategies have been published in the literature. In this paper, we propose one of the first defense strategies against multiple types of adversarial videos for video recognition. The proposed method, referred to as MultiBN, performs adversarial training on multiple adversarial video types using multiple independent batch normalization (BN) layers with a learning-based BN selection module. With a multiple BN structure, each BN brach is responsible for learning the distribution of a single perturbation type and thus provides more precise distribution estimations. This mechanism benefits dealing with multiple perturbation types. The BN selection module detects the attack type of an input video and sends it to the corresponding BN branch, making MultiBN fully automatic and allowing end-to-end training. Compared to present adversarial training approaches, the proposed MultiBN exhibits stronger multi-perturbation robustness against different and even unforeseen adversarial video types, ranging from Lp-bounded attacks and physically realizable attacks. This holds true on different datasets and target models. Moreover, we conduct an extensive analysis to study the properties of the multiple BN structure.
The task of object segmentation in videos is usually accomplished by processing appearance and motion information separately using standard 2D convolutional networks, followed by a learned fusion of the two sources of information. On the other hand, 3D convolutional networks have been successfully applied for video classification tasks, but have not been leveraged as effectively to problems involving dense per-pixel interpretation of videos compared to their 2D convolutional counterparts and lag behind the aforementioned networks in terms of performance. In this work, we show that 3D CNNs can be effectively applied to dense video prediction tasks such as salient object segmentation. We propose a simple yet effective encoder-decoder network architecture consisting entirely of 3D convolutions that can be trained end-to-end using a standard cross-entropy loss. To this end, we leverage an efficient 3D encoder, and propose a 3D decoder architecture, that comprises novel 3D Global Convolution layers and 3D Refinement modules. Our approach outperforms existing state-of-the-arts by a large margin on the DAVIS'16 Unsupervised, FBMS and ViSal dataset benchmarks in addition to being faster, thus showing that our architecture can efficiently learn expressive spatio-temporal features and produce high quality video segmentation masks. We have made our code and trained models publicly available at https://github.com/sabarim/3DC-Seg.
This paper proposes a novel model for video generation and especially makes the attempt to deal with the problem of video generation from text descriptions, i.e., synthesizing realistic videos conditioned on given texts. Existing video generation methods cannot be easily adapted to handle this task well, due to the frame discontinuity issue and their text-free generation schemes. To address these problems, we propose a recurrent deconvolutional generative adversarial network (RD-GAN), which includes a recurrent deconvolutional network (RDN) as the generator and a 3D convolutional neural network (3D-CNN) as the discriminator. The RDN is a deconvolutional version of conventional recurrent neural network, which can well model the long-range temporal dependency of generated video frames and make good use of conditional information. The proposed model can be jointly trained by pushing the RDN to generate realistic videos so that the 3D-CNN cannot distinguish them from real ones. We apply the proposed RD-GAN to a series of tasks including conventional video generation, conditional video generation, video prediction and video classification, and demonstrate its effectiveness by achieving well performance.
This technical report summarizes submissions and compiles from Actor-Action video classification challenge held as a final project in CSC 249/449 Machine Vision course (Spring 2020) at University of Rochester
We consider two less-emphasized temporal properties of video: 1. Temporal cues are fine-grained; 2. Temporal modeling needs reasoning. To tackle both problems at once, we exploit approximated bilinear modules (ABMs) for temporal modeling. There are two main points making the modules effective: two-layer MLPs can be seen as a constraint approximation of bilinear operations, thus can be used to construct deep ABMs in existing CNNs while reusing pretrained parameters; frame features can be divided into static and dynamic parts because of visual repetition in adjacent frames, which enables temporal modeling to be more efficient. Multiple ABM variants and implementations are investigated, from high performance to high efficiency. Specifically, we show how two-layer subnets in CNNs can be converted to temporal bilinear modules by adding an auxiliary-branch. Besides, we introduce snippet sampling and shifting inference to boost sparse-frame video classification performance. Extensive ablation studies are conducted to show the effectiveness of proposed techniques. Our models can outperform most state-of-the-art methods on Something-Something v1 and v2 datasets without Kinetics pretraining, and are also competitive on other YouTube-like action recognition datasets. Our code is available on https://github.com/zhuxinqimac/abm-pytorch.
Convolutional operations have two limitations: (1) do not explicitly model where to focus as the same filter is applied to all the positions, and (2) are unsuitable for modeling long-range dependencies as they only operate on a small neighborhood. While both limitations can be alleviated by attention operations, many design choices remain to be determined to use attention, especially when applying attention to videos. Towards a principled way of applying attention to videos, we address the task of spatiotemporal attention cell search. We propose a novel search space for spatiotemporal attention cells, which allows the search algorithm to flexibly explore various design choices in the cell. The discovered attention cells can be seamlessly inserted into existing backbone networks, e.g., I3D or S3D, and improve video classification accuracy by more than 2% on both Kinetics-600 and MiT datasets. The discovered attention cells outperform non-local blocks on both datasets, and demonstrate strong generalization across different modalities, backbones, and datasets. Inserting our attention cells into I3D-R50 yields state-of-the-art performance on both datasets.
Despite the recent advances in video classification, progress in spatio-temporal action recognition has lagged behind. A major contributing factor has been the prohibitive cost of annotating videos frame-by-frame. In this paper, we present a spatio-temporal action recognition model that is trained with only video-level labels, which are significantly easier to annotate. Our method leverages per-frame person detectors which have been trained on large image datasets within a Multiple Instance Learning framework. We show how we can apply our method in cases where the standard Multiple Instance Learning assumption, that each bag contains at least one instance with the specified label, is invalid using a novel probabilistic variant of MIL where we estimate the uncertainty of each prediction. Furthermore, we report the first weakly-supervised results on the AVA dataset and state-of-the-art results among weakly-supervised methods on UCF101-24.
Motion plays a crucial role in understanding videos and most state-of-the-art neural models for video classification incorporate motion information typically using optical flows extracted by a separate off-the-shelf method. As the frame-by-frame optical flows require heavy computation, incorporating motion information has remained a major computational bottleneck for video understanding. In this work, we replace external and heavy computation of optical flows with internal and light-weight learning of motion features. We propose a trainable neural module, dubbed MotionSqueeze, for effective motion feature extraction. Inserted in the middle of any neural network, it learns to establish correspondences across frames and convert them into motion features, which are readily fed to the next downstream layer for better prediction. We demonstrate that the proposed method provides a significant gain on four standard benchmarks for action recognition with only a small amount of additional cost, outperforming the state of the art on Something-Something-V1&V2 datasets.
Convolutional Neural Networks (CNNs) model long-range dependencies by deeply stacking convolution operations with small window sizes, which makes the optimizations difficult. This paper presents region-based non-local (RNL) operations as a family of self-attention mechanisms, which can directly capture long-range dependencies without using a deep stack of local operations. Given an intermediate feature map, our method recalibrates the feature at a position by aggregating the information from the neighboring regions of all positions. By combining a channel attention module with the proposed RNL, we design an attention chain, which can be integrated into the off-the-shelf CNNs for end-to-end training. We evaluate our method on two video classification benchmarks. The experimental results of our method outperform other attention mechanisms, and we achieve state-of-the-art performance on the Something-Something V1 dataset.
Geological parameterization enables the representation of geomodels in terms of a relatively small set of variables. Parameterization is therefore very useful in the context of data assimilation and uncertainty quantification. In this study, a deep-learning-based geological parameterization algorithm, CNN-PCA, is developed for complex 3D geomodels. CNN-PCA entails the use of convolutional neural networks as a post-processor for the low-dimensional principal component analysis representation of a geomodel. The 3D treatments presented here differ somewhat from those used in the 2D CNN-PCA procedure. Specifically, we introduce a new supervised-learning-based reconstruction loss, which is used in combination with style loss and hard data loss. The style loss uses features extracted from a 3D CNN pretrained for video classification. The 3D CNN-PCA algorithm is applied for the generation of conditional 3D realizations, defined on $60\times60\times40$ grids, for three geological scenarios (binary and bimodal channelized systems, and a three-facies channel-levee-mud system). CNN-PCA realizations are shown to exhibit geological features that are visually consistent with reference models generated using object-based methods. Statistics of flow responses ($\text{P}_{10}$, $\text{P}_{50}$, $\text{P}_{90}$ percentile results) for test sets of 3D CNN-PCA models are shown to be in consistent agreement with those from reference geomodels. Lastly, CNN-PCA is successfully applied for history matching with ESMDA for the bimodal channelized system.
Few-shot learning aims to recognize novel classes from a few examples. Although significant progress has been made in the image domain, few-shot video classification is relatively unexplored. We argue that previous methods underestimate the importance of video feature learning and propose to learn spatiotemporal features using a 3D CNN. Proposing a two-stage approach that learns video features on base classes followed by fine-tuning the classifiers on novel classes, we show that this simple baseline approach outperforms prior few-shot video classification methods by over 20 points on existing benchmarks. To circumvent the need of labeled examples, we present two novel approaches that yield further improvement. First, we leverage tag-labeled videos from a large dataset using tag retrieval followed by selecting the best clips with visual similarities. Second, we learn generative adversarial networks that generate video features of novel classes from their semantic embeddings. Moreover, we find existing benchmarks are limited because they only focus on 5 novel classes in each testing episode and introduce more realistic benchmarks by involving more novel classes, i.e. few-shot learning, as well as a mixture of novel and base classes, i.e. generalized few-shot learning. The experimental results show that our retrieval and feature generation approach significantly outperform the baseline approach on the new benchmarks.
Temporal convolution has been widely used for video classification. However, it is performed on spatio-temporal contexts in a limited view, which often weakens its capacity of learning video representation. To alleviate this problem, we propose a concise and novel SmallBig network, with the cooperation of small and big views. For the current time step, the small view branch is used to learn the core semantics, while the big view branch is used to capture the contextual semantics. Unlike traditional temporal convolution, the big view branch can provide the small view branch with the most activated video features from a broader 3D receptive field. Via aggregating such big-view contexts, the small view branch can learn more robust and discriminative spatio-temporal representations for video classification. Furthermore, we propose to share convolution in the small and big view branch, which improves model compactness as well as alleviates overfitting. As a result, our SmallBigNet achieves a comparable model size like 2D CNNs, while boosting accuracy like 3D CNNs. We conduct extensive experiments on the large-scale video benchmarks, e.g., Kinetics400, Something-Something V1 and V2. Our SmallBig network outperforms a number of recent state-of-the-art approaches, in terms of accuracy and/or efficiency. The codes and models will be available on https://github.com/xhl-video/SmallBigNet.
This work introduces pyramidal convolution (PyConv), which is capable of processing the input at multiple filter scales. PyConv contains a pyramid of kernels, where each level involves different types of filters with varying size and depth, which are able to capture different levels of details in the scene. On top of these improved recognition capabilities, PyConv is also efficient and, with our formulation, it does not increase the computational cost and parameters compared to standard convolution. Moreover, it is very flexible and extensible, providing a large space of potential network architectures for different applications. PyConv has the potential to impact nearly every computer vision task and, in this work, we present different architectures based on PyConv for four main tasks on visual recognition: image classification, video action classification/recognition, object detection and semantic image segmentation/parsing. Our approach shows significant improvements over all these core tasks in comparison with the baselines. For instance, on image recognition, our 50-layers network outperforms in terms of recognition performance on ImageNet dataset its counterpart baseline ResNet with 152 layers, while having 2.39 times less parameters, 2.52 times lower computational complexity and more than 3 times less layers. On image segmentation, our novel framework sets a new state-of-the-art on the challenging ADE20K benchmark for scene parsing. Code is available at: https://github.com/iduta/pyconv
Generalizing over temporal variations is a prerequisite for effective action recognition in videos. Despite significant advances in deep neural networks, it remains a challenge to focus on short-term discriminative motions in relation to the overall performance of an action. We address this challenge by allowing some flexibility in discovering relevant spatio-temporal features. We introduce Squeeze and Recursion Temporal Gates (SRTG), an approach that favors inputs with similar activations with potential temporal variations. We implement this idea with a novel CNN block that uses an LSTM to encapsulate feature dynamics, in conjunction with a temporal gate that is responsible for evaluating the consistency of the discovered dynamics and the modeled features. We show consistent improvement when using SRTG blocks, with only a minimal increase in the number of GFLOPs. On Kinetics-700, we perform on par with current state-of-the-art models, and outperform these on HACS, Moments in Time, UCF-101 and HMDB-51.
With the advent of large-scale multimodal video datasets, especially sequences with audio or transcribed speech, there has been a growing interest in self-supervised learning of video representations. Most prior work formulates the objective as a contrastive metric learning problem between the modalities. To enable effective learning, however, these strategies require a careful selection of positive and negative samples often combined with hand-designed curriculum policies. In this work we remove the need for negative sampling by taking a generative modeling approach that poses the objective as a translation problem between modalities. Such a formulation allows us to tackle a wide variety of downstream video understanding tasks by means of a single unified framework, without the need for large batches of negative samples common in contrastive metric learning. We experiment with the large-scale HowTo100M dataset for training, and report performance gains over the state-of-the-art on several downstream tasks including video classification (EPIC-Kitchens), question answering (TVQA), captioning (TVC, YouCook2, and MSR-VTT), and text-based clip retrieval (YouCook2 and MSR-VTT).
We propose a novel application of Transfer Learning to classify video-frame sequences over multiple classes. This is a pre-weighted model that does not require to train a fresh CNN. This representation is achieved with the advent of "deep neural network" (DNN), which is being studied these days by many researchers. We utilize the classical approaches for video classification task using object detection techniques for comparison, such as "Google Video Intelligence API" and this study will run experiments as to how those architectures would perform in foggy or rainy weather conditions. Experimental evaluation on video collections shows that the new proposed classifier achieves superior performance over existing solutions.
Face anti-spoofing has become an increasingly important and critical security feature for authentication systems, due to rampant and easily launchable presentation attacks. Addressing the shortage of multi-modal face dataset, CASIA recently released the largest up-to-date CASIA-SURF Cross-ethnicity Face Anti-spoofing(CeFA) dataset, covering 3 ethnicities, 3 modalities, 1607 subjects, and 2D plus 3D attack types in four protocols, and focusing on the challenge of improving the generalization capability of face anti-spoofing in cross-ethnicity and multi-modal continuous data. In this paper, we propose a novel pipeline-based multi-stream CNN architecture called PipeNet for multi-modal face anti-spoofing. Unlike previous works, Selective Modal Pipeline (SMP) is designed to enable a customized pipeline for each data modality to take full advantage of multi-modal data. Limited Frame Vote (LFV) is designed to ensure stable and accurate prediction for video classification. The proposed method wins the third place in the final ranking of Chalearn Multi-modal Cross-ethnicity Face Anti-spoofing Recognition Challenge@CVPR2020. Our final submission achieves the Average Classification Error Rate (ACER) of 2.21 with Standard Deviation of 1.26 on the test set.
Classification of new class entities requires collecting and annotating hundreds or thousands of samples that is often prohibitively costly. Few-shot learning suggests learning to classify new classes using just a few examples. Only a small number of studies address the challenge of few-shot learning on spatio-temporal patterns such as videos. In this paper, we present the Temporal Aware Embedding Network (TAEN) for few-shot action recognition, that learns to represent actions, in a metric space as a trajectory, conveying both short term semantics and longer term connectivity between action parts. We demonstrate the effectiveness of TAEN on two few shot tasks, video classification and temporal action detection and evaluate our method on the Kinetics-400 and on ActivityNet 1.2 few-shot benchmarks. With training of just a few fully connected layers we reach comparable results to prior art on both few shot video classification and temporal detection tasks, while reaching state-of-the-art in certain scenarios.
How can we collect and use a video dataset to further improve spatiotemporal 3D Convolutional Neural Networks (3D CNNs)? In order to positively answer this open question in video recognition, we have conducted an exploration study using a couple of large-scale video datasets and 3D CNNs. In the early era of deep neural networks, 2D CNNs have been better than 3D CNNs in the context of video recognition. Recent studies revealed that 3D CNNs can outperform 2D CNNs trained on a large-scale video dataset. However, we heavily rely on architecture exploration instead of dataset consideration. Therefore, in the present paper, we conduct exploration study in order to improve spatiotemporal 3D CNNs as follows: (i) Recently proposed large-scale video datasets help improve spatiotemporal 3D CNNs in terms of video classification accuracy. We reveal that a carefully annotated dataset (e.g., Kinetics-700) effectively pre-trains a video representation for a video classification task. (ii) We confirm the relationships between #category/#instance and video classification accuracy. The results show that #category should initially be fixed, and then #instance is increased on a video dataset in case of dataset construction. (iii) In order to practically extend a video dataset, we simply concatenate publicly available datasets, such as Kinetics-700 and Moments in Time (MiT) datasets. Compared with Kinetics-700 pre-training, we further enhance spatiotemporal 3D CNNs with the merged dataset, e.g., +0.9, +3.4, and +1.1 on UCF-101, HMDB-51, and ActivityNet datasets, respectively, in terms of fine-tuning. (iv) In terms of recognition architecture, the Kinetics-700 and merged dataset pre-trained models increase the recognition performance to 200 layers with the Residual Network (ResNet), while the Kinetics-400 pre-trained model cannot successfully optimize the 200-layer architecture.
This paper presents X3D, a family of efficient video networks that progressively expand a tiny 2D image classification architecture along multiple network axes, in space, time, width and depth. Inspired by feature selection methods in machine learning, a simple stepwise network expansion approach is employed that expands a single axis in each step, such that good accuracy to complexity trade-off is achieved. To expand X3D to a specific target complexity, we perform progressive forward expansion followed by backward contraction. X3D achieves state-of-the-art performance while requiring 4.8x and 5.5x fewer multiply-adds and parameters for similar accuracy as previous work. Our most surprising finding is that networks with high spatiotemporal resolution can perform well, while being extremely light in terms of network width and parameters. We report competitive accuracy at unprecedented efficiency on video classification and detection benchmarks. Code will be available at: https://github.com/facebookresearch/SlowFast