3D Attention Mechanism for Fine-Grained Classification of Table Tennis Strokes Using a Twin Spatio-Temporal Convolutional Neural Networks

Pierre-Etienne Martin, Jenny Benois-Pineau, Renaud Péteri, Julien Morlier

Responsive image

Auto-TLDR; Attentional Blocks for Action Recognition in Table Tennis Strokes

Slides Poster

The paper addresses the problem of recognition of actions in video with low inter-class variability such as Table Tennis strokes. Two stream, "twin" convolutional neural networks are used with 3D convolutions both on RGB data and optical flow. Actions are recognized by classification of temporal windows. We introduce 3D attention modules and examine their impact on classification efficiency. In the context of the study of sportsmen performances, a corpus of the particular actions of table tennis strokes is considered. The use of attention blocks in the network speeds up the training step and improves the classification scores up to 5% with our twin model. We visualize the impact on the obtained features and notice correlation between attention and player movements and position. Score comparison of state-of-the-art action classification method and proposed approach with attentional blocks is performed on the corpus. Proposed model with attention blocks outperforms previous model without them and our baseline.

Similar papers

Self-Supervised Joint Encoding of Motion and Appearance for First Person Action Recognition

Mirco Planamente, Andrea Bottino, Barbara Caputo

Responsive image

Auto-TLDR; A Single Stream Architecture for Egocentric Action Recognition from the First-Person Point of View

Slides Poster Similar

Wearable cameras are becoming more and more popular in several applications, increasing the interest of the research community in developing approaches for recognizing actions from the first-person point of view. An open challenge in egocentric action recognition is that videos lack detailed information about the main actor's pose and thus tend to record only parts of the movement when focusing on manipulation tasks. Thus, the amount of information about the action itself is limited, making crucial the understanding of the manipulated objects and their context. Many previous works addressed this issue with two-stream architectures, where one stream is dedicated to modeling the appearance of objects involved in the action, and another to extracting motion features from optical flow. In this paper, we argue that learning features jointly from these two information channels is beneficial to capture the spatio-temporal correlations between the two better. To this end, we propose a single stream architecture able to do so, thanks to the addition of a self-supervised block that uses a pretext motion prediction task to intertwine motion and appearance knowledge. Experiments on several publicly available databases show the power of our approach.

Region-Based Non-Local Operation for Video Classification

Guoxi Huang, Adrian Bors

Responsive image

Auto-TLDR; Regional-based Non-Local Operation for Deep Self-Attention in Convolutional Neural Networks

Slides Poster Similar

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 operation (RNL), a family of self-attention mechanisms, which can directly capture long-range dependencies without a deep stack of local operations. Given an intermediate feature map, our method recalibrates the feature at a position by aggregating 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 off-the-shelf CNNs for end-to-end training. We evaluate our method on two video classification benchmarks. The experimental result of our method outperforms other attention mechanisms, and we achieve state-of-the-art performance on Something-Something V1.

Single View Learning in Action Recognition

Gaurvi Goyal, Nicoletta Noceti, Francesca Odone

Responsive image

Auto-TLDR; Cross-View Action Recognition Using Domain Adaptation for Knowledge Transfer

Slides Poster Similar

Viewpoint is an essential aspect of how an action is visually perceived, with the motion appearing substantially different for some viewpoint pairs. Data driven action recognition algorithms compensate for this by including a variety of viewpoints in their training data, adding to the cost of data acquisition as well as training. We propose a novel methodology that leverages deeply pretrained features to learn actions from a single viewpoint using domain adaptation for knowledge transfer. We demonstrate the effectiveness of this pipeline on 3 different datasets: IXMAS, MoCA and NTU RGBD+, and compare with both classical and deep learning methods. Our method requires low training data and demonstrates unparalleled cross-view action recognition accuracies for single view learning.

Learnable Higher-Order Representation for Action Recognition

Jie Shao, Xiangyang Xue

Responsive image

Auto-TLDR; Learningable Higher-Order Operations for Spatiotemporal Dynamics in Video Recognition

Similar

Capturing spatiotemporal dynamics is an essential topic in video recognition. In this paper, we present learnable higher-order operations as a generic family of building blocks for capturing spatiotemporal dynamics from RGB input video space. Similar to higher-order functions, the weights of higher-order operations are themselves derived from the data with learnable parameters. Classical architectures such as residual learning and network-in-network are first-order operations where weights are directly learned from the data. Higher-order operations make it easier to capture context-sensitive patterns, such as motion. Self-attention models are also higher-order operations, but the attention weights are mostly computed from an affine operation or dot product. The learnable higher-order operations can be more generic and flexible. Experimentally, we show that on the task of video recognition, our higher-order models can achieve results on par with or better than the existing state-of-the-art methods on Something-Something (V1 and V2), Kinetics and Charades datasets.

TinyVIRAT: Low-Resolution Video Action Recognition

Ugur Demir, Yogesh Rawat, Mubarak Shah

Responsive image

Auto-TLDR; TinyVIRAT: A Progressive Generative Approach for Action Recognition in Videos

Slides Poster Similar

The existing research in action recognition is mostly focused on high-quality videos where the action is distinctly visible. In real-world surveillance environments, the actions in videos are captured at a wide range of resolutions. Most activities occur at a distance with a small resolution and recognizing such activities is a challenging problem. In this work, we focus on recognizing tiny actions in videos. We introduce a benchmark dataset, TinyVIRAT, which contains natural low-resolution activities. The actions in TinyVIRAT videos have multiple labels and they are extracted from surveillance videos which makes them realistic and more challenging. We propose a novel method for recognizing tiny actions in videos which utilizes a progressive generative approach to improve the quality of low-resolution actions. The proposed method also consists of a weakly trained attention mechanism which helps in focusing on the activity regions in the video. We perform extensive experiments to benchmark the proposed TinyVIRAT dataset and observe that the proposed method significantly improves the action recognition performance over baselines. We also evaluate the proposed approach on synthetically resized action recognition datasets and achieve state-of-the-art results when compared with existing methods. The dataset and code will be publicly available.

RWF-2000: An Open Large Scale Video Database for Violence Detection

Ming Cheng, Kunjing Cai, Ming Li

Responsive image

Auto-TLDR; Flow Gated Network for Violence Detection in Surveillance Cameras

Slides Poster Similar

In recent years, surveillance cameras are widely deployed in public places, and the general crime rate has been reduced significantly due to these ubiquitous devices. Usually, these cameras provide cues and evidence after crimes were conducted, while they are rarely used to prevent or stop criminal activities in time. It is both time and labor consuming to manually monitor a large amount of video data from surveillance cameras. Therefore, automatically recognizing violent behaviors from video signals becomes essential. In this paper, we summarize several existing video datasets for violence detection and propose a new video dataset with 2,000 videos all captured by surveillance cameras in real-world scenes. Also, we present a new method that utilizes both the merits of 3D-CNNs and optical flow, namely Flow Gated Network. The proposed approach obtains an accuracy of 87.25% on the test set of our proposed RWF-2000 database. The proposed database and source codes of this paper are currently open to access.

MFI: Multi-Range Feature Interchange for Video Action Recognition

Sikai Bai, Qi Wang, Xuelong Li

Responsive image

Auto-TLDR; Multi-range Feature Interchange Network for Action Recognition in Videos

Slides Poster Similar

Short-range motion features and long-range dependencies are two complementary and vital cues for action recognition in videos, but it remains unclear how to efficiently and effectively extract these two features. In this paper, we propose a novel network to capture these two features in a unified 2D framework. Specifically, we first construct a Short-range Temporal Interchange (STI) block, which contains a Channels-wise Temporal Interchange (CTI) module for encoding short-range motion features. Then a Graph-based Regional Interchange (GRI) module is built to present long-range dependencies using graph convolution. Finally, we replace original bottleneck blocks in the ResNet with STI blocks and insert several GRI modules between STI blocks, to form a Multi-range Feature Interchange (MFI) Network. Practically, extensive experiments are conducted on three action recognition datasets (i.e., Something-Something V1, HMDB51, and UCF101), which demonstrate that the proposed MFI network achieves impressive results with very limited computing cost.

Motion Complementary Network for Efficient Action Recognition

Ke Cheng, Yifan Zhang, Chenghua Li, Jian Cheng, Hanqing Lu

Responsive image

Auto-TLDR; Efficient Motion Complementary Network for Action Recognition

Slides Poster Similar

Both two-stream ConvNet and 3D ConvNet are widely used in action recognition. However, both methods are not efficient for deployment: calculating optical flow is very slow, while 3D convolution is computationally expensive. Our key insight is that the motion information from optical flow maps is complementary to the motion information from 3D ConvNet. Instead of simply combining these two methods, we propose two novel techniques to enhance the performance with less computational cost: \textit{fixed-motion-accumulation} and \textit{balanced-motion-policy}. With these two techniques, we propose a novel framework called Efficient Motion Complementary Network(EMC-Net) that enjoys both high efficiency and high performance. We conduct extensive experiments on Kinetics, UCF101, and Jester datasets. We achieve notably higher performance while consuming 4.7$\times$ less computation than I3D, 11.6$\times$ less computation than ECO, 17.8$\times$ less computation than R(2+1)D. On Kinetics dataset, we achieve 2.6\% better performance than the recent proposed TSM with 1.4$\times$ fewer FLOPs and 10ms faster on K80 GPU.

Channel-Wise Dense Connection Graph Convolutional Network for Skeleton-Based Action Recognition

Michael Lao Banteng, Zhiyong Wu

Responsive image

Auto-TLDR; Two-stream channel-wise dense connection GCN for human action recognition

Slides Poster Similar

Skeleton-based action recognition task has drawn much attention for many years. Graph Convolutional Network (GCN) has proved its effectiveness in this task. However, how to improve the model's robustness to different human actions and how to make effective use of features produced by the network are main topics needed to be further explored. Human actions are time series sequence, meaning that temporal information is a key factor to model the representation of data. The ranges of body parts involved in small actions (e.g. raise a glass or shake head) and big actions (e.g. walking or jumping) are diverse. It's crucial for the model to generate and utilize more features that can be adaptive to a wider range of actions. Furthermore, feature channels are specific with the action class, the model needs to weigh their importance and pay attention to more related ones. To address these problems, in this work, we propose a two-stream channel-wise dense connection GCN (2s-CDGCN). Specifically, the skeleton data was extracted and processed into spatial and temporal information for better feature representation. A channel-wise attention module was used to select and emphasize the more useful features generated by the network. Moreover, to ensure maximum information flow, dense connection was introduced to the network structure, which enables the network to reuse the skeleton features and generate more information adaptive and related to different human actions. Our model has shown its ability to improve the accuracy of human action recognition task on two large datasets, NTU-RGB+D and Kinetics. Extensive evaluations were conducted to prove the effectiveness of our model.

MixTConv: Mixed Temporal Convolutional Kernels for Efficient Action Recognition

Kaiyu Shan, Yongtao Wang, Zhi Tang, Ying Chen, Yangyan Li

Responsive image

Auto-TLDR; Mixed Temporal Convolution for Action Recognition

Slides Poster Similar

To efficiently extract spatiotemporal features of video for action recognition, most state-of-the-art methods integrate 1D temporal convolution into a conventional 2D CNN backbone. However, they all exploit 1D temporal convolution of fixed kernel size (i.e., 3) in the network building block, thus have suboptimal temporal modeling capability to handle both long term and short-term actions. To address this problem, we first investigate the impacts of different kernel sizes for the 1D temporal convolutional filters. Then, we propose a simple yet efficient operation called Mixed Temporal Convolution (MixTConv) in methodology, which consists of multiple depthwise 1D convolutional filters with different kernel sizes. By plugging MixTConv into the conventional 2D CNN backbone ResNet-50, we further propose an efficient and effective network architecture named MSTNet for action recognition, and achieve state-of-the-art results on multiple large-scale benchmarks.

RMS-Net: Regression and Masking for Soccer Event Spotting

Matteo Tomei, Lorenzo Baraldi, Simone Calderara, Simone Bronzin, Rita Cucchiara

Responsive image

Auto-TLDR; An Action Spotting Network for Soccer Videos

Slides Poster Similar

The recently proposed action spotting task consists in finding the exact timestamp in which an event occurs. This task fits particularly well for soccer videos, where events correspond to salient actions strictly defined by soccer rules (a goal occurs when the ball crosses the goal line). In this paper, we devise a lightweight and modular network for action spotting, which can simultaneously predict the event label and its temporal offset using the same underlying features. We enrich our model with two training strategies: the first one for data balancing and uniform sampling, the second for masking ambiguous frames and keeping the most discriminative visual cues. When tested on the SoccerNet dataset and using standard features, our full proposal exceeds the current state of the art by 3 Average-mAP points. Additionally, it reaches a gain of more than 10 Average-mAP points on the test set when fine-tuned in combination with a strong 2D backbone.

SCA Net: Sparse Channel Attention Module for Action Recognition

Hang Song, Yonghong Song, Yuanlin Zhang

Responsive image

Auto-TLDR; SCA Net: Efficient Group Convolution for Sparse Channel Attention

Slides Poster Similar

Channel attention has shown its great performance recently when it was incorporated into deep convolutional neural networks. However, existing methods usually require extensive computing resources due to their involuted structure, which is hard for 3D CNNs to take full advantage of. In this paper, a lightweight sparse channel attention (SCA) module implemented by efficient group convolution is proposed, which adopts the idea of sparse channel connection and involves much less parameters but brings clear performance gain. Meanwhile, to solve the lack of local channel interaction brought by group convolution, a dominant function called Aggregate-Shuffle-Diverge (ASD) is leveraged to enhance information flow over each group with no additional parameters. We also adjust the existing mainstream 3D CNNs by employing 3D convolution factorization, so as to further reduce the parameters. Our SCA module can be flexibly incorporated into most existing 3D CNNs, all of which can achieve a perfect trade-off between performance and complexity on action recognition task with factorized I3D or 3D ResNet backbone networks. The experimental results also indicate that the resulting network, namely, SCA Net can achieve an outstanding performance on UCF-101 and HMDB-51 datasets.

Attention-Driven Body Pose Encoding for Human Activity Recognition

Bappaditya Debnath, Swagat Kumar, Marry O'Brien, Ardhendu Behera

Responsive image

Auto-TLDR; Attention-based Body Pose Encoding for Human Activity Recognition

Slides Poster Similar

This article proposes a novel attention-based body pose encoding for human activity recognition. Most of the existing human activity recognition approaches based on 3D pose data often enrich the input data using additional handcrafted representations such as velocity, super normal vectors, pairwise relations, and so on. The enriched data complements the 3D body joint position data and improves the model performance. In this paper, we propose a novel approach that learns enhanced feature representations from a given sequence of 3D body joints. To achieve this, the approach exploits two body pose streams: 1) a spatial stream which encodes the spatial relationship between various body joints at each time point to learn spatial structure involving the spatial distribution of different body joints 2) a temporal stream that learns the temporal variation of individual body joints over the entire sequence duration to present a temporally enhanced representation. Afterwards, these two pose streams are fused with a multi-head attention mechanism. We also capture the contextual information from the RGB video stream using a deep Convolutional Neural Network (CNN) model combined with a multi-head attention and a bidirectional Long Short-Term Memory (LSTM) network. Finally, the RGB video stream is combined with the fused body pose stream to give a novel end-to-end deep model for effective human activity recognition. The proposed model is evaluated on three datasets including the challenging NTU-RGBD dataset and achieves state-of-the-art results.

Modeling Long-Term Interactions to Enhance Action Recognition

Alejandro Cartas, Petia Radeva, Mariella Dimiccoli

Responsive image

Auto-TLDR; A Hierarchical Long Short-Term Memory Network for Action Recognition in Egocentric Videos

Slides Poster Similar

In this paper, we propose a new approach to understand actions in egocentric videos that exploit the semantics of object interactions at both frame and temporal levels. At the frame level, we use a region-based approach that takes as input a primary region roughly corresponding to the user hands and a set of secondary regions potentially corresponding to the interacting objects and calculates the action score through a CNN formulation. This information is then fed to a Hierarchical Long Short-Term Memory Network (HLSTM) that captures temporal dependencies between actions within and across shots. Ablation studies thoroughly validate the proposed approach, showing in particular that both levels of the HLSTM architecture contribute to performance improvement. Furthermore, quantitative comparisons show that the proposed approach outperforms the state-of-the-art in terms of action recognition on standard benchmarks, without relying on motion information.

Flow-Guided Spatial Attention Tracking for Egocentric Activity Recognition

Tianshan Liu, Kin-Man Lam

Responsive image

Auto-TLDR; flow-guided spatial attention tracking for egocentric activity recognition

Slides Poster Similar

The popularity of wearable cameras has opened up a new dimension for egocentric activity recognition. While some methods introduce attention mechanisms into deep learning networks to capture fine-grained hand-object interactions, they often neglect exploring the spatio-temporal relationships. Generating spatial attention, without adequately exploiting temporal consistency, will result in potentially sub-optimal performance in the video-based task. In this paper, we propose a flow-guided spatial attention tracking (F-SAT) module, which is based on enhancing motion patterns and inter-frame information, to highlight the discriminative features from regions of interest across a video sequence. A new form of input, namely the optical-flow volume, is presented to provide informative cues from moving parts for spatial attention tracking. The proposed F-SAT module is deployed to a two-branch-based deep architecture, which fuses complementary information for egocentric activity recognition. Experimental results on three egocentric activity benchmarks show that the proposed method achieves state-of-the-art performance.

Developing Motion Code Embedding for Action Recognition in Videos

Maxat Alibayev, David Andrea Paulius, Yu Sun

Responsive image

Auto-TLDR; Motion Embedding via Motion Codes for Action Recognition

Slides Poster Similar

We propose a motion embedding strategy via the motion codes that is a vectorized representation of motions based on their salient mechanical attributes. We show that our motion codes can provide robust motion representation. We train a deep neural network model that learns to embed demonstration videos into motion codes. We integrate the extracted features from the motion embedding model into the current state-of-the-art action recognition model. The obtained model achieved higher accuracy than the baseline on a verb classification task from egocentric videos in EPIC-KITCHENS dataset.

AttendAffectNet: Self-Attention Based Networks for Predicting Affective Responses from Movies

Thi Phuong Thao Ha, Bt Balamurali, Herremans Dorien, Roig Gemma

Responsive image

Auto-TLDR; AttendAffectNet: A Self-Attention Based Network for Emotion Prediction from Movies

Slides Poster Similar

In this work, we propose different variants of the self-attention based network for emotion prediction from movies, which we call AttendAffectNet. We take both audio and video into account and incorporate the relation among multiple modalities by applying self-attention mechanism in a novel manner into the extracted features for emotion prediction. We compare it to the typically temporal integration of the self-attention based model, which in our case, allows to capture the relation of temporal representations of the movie while considering the sequential dependencies of emotion responses. We demonstrate the effectiveness of our proposed architectures on the extended COGNIMUSE dataset [1], [2] and the MediaEval 2016 Emotional Impact of Movies Task [3], which consist of movies with emotion annotations. Our results show that applying the self-attention mechanism on the different audio-visual features, rather than in the time domain, is more effective for emotion prediction. Our approach is also proven to outperform state-of-the-art models for emotion prediction.

Improved Residual Networks for Image and Video Recognition

Ionut Cosmin Duta, Li Liu, Fan Zhu, Ling Shao

Responsive image

Auto-TLDR; Residual Networks for Deep Learning

Slides Poster Similar

Residual networks (ResNets) represent a powerful type of convolutional neural network (CNN) architecture, widely adopted and used in various tasks. In this work we propose an improved version of ResNets. Our proposed improvements address all three main components of a ResNet: the flow of information through the network layers, the residual building block, and the projection shortcut. We are able to show consistent improvements in accuracy and learning convergence over the baseline. For instance, on ImageNet dataset, using the ResNet with 50 layers, for top-1 accuracy we can report a 1.19% improvement over the baseline in one setting and around 2% boost in another. Importantly, these improvements are obtained without increasing the model complexity. Our proposed approach allows us to train extremely deep networks, while the baseline shows severe optimization issues. We report results on three tasks over six datasets: image classification (ImageNet, CIFAR-10 and CIFAR-100), object detection (COCO) and video action recognition (Kinetics-400 and Something-Something-v2). In the deep learning era, we establish a new milestone for the depth of a CNN. We successfully train a 404-layer deep CNN on the ImageNet dataset and a 3002-layer network on CIFAR-10 and CIFAR-100, while the baseline is not able to converge at such extreme depths. Code is available at: https://github.com/iduta/iresnet

What and How? Jointly Forecasting Human Action and Pose

Yanjun Zhu, Yanxia Zhang, Qiong Liu, Andreas Girgensohn

Responsive image

Auto-TLDR; Forecasting Human Actions and Motion Trajectories with Joint Action Classification and Pose Regression

Slides Poster Similar

Forecasting human actions and motion trajectories addresses the problem of predicting what a person is going to do next and how they will perform it. This is crucial in a wide range of applications such as assisted living and future co-robotic settings. We propose to simultaneously learn actions and action-related human motion dynamics, while existing works perform them independently. In this paper, we present a method to jointly forecast categories of human action and the pose of skeletal joints in the hope that the two tasks can help each other. As a result, our system can predict not only the future actions but also the motion trajectories that will result. To achieve this, we define a task of joint action classification and pose regression. We employ a sequence to sequence encoder-decoder model combined with multi-task learning to forecast future actions and poses progressively before the action happens. Experimental results on two public datasets, IkeaDB and OAD, demonstrate the effectiveness of the proposed method.

2D Deep Video Capsule Network with Temporal Shift for Action Recognition

Théo Voillemin, Hazem Wannous, Jean-Philippe Vandeborre

Responsive image

Auto-TLDR; Temporal Shift Module over Capsule Network for Action Recognition in Continuous Videos

Slides Similar

Action recognition in continuous video streams is a growing field since the past few years. Deep learning techniques and in particular Convolutional Neural Networks (CNNs) achieved good results in this topic. However, intrinsic CNNs limitations begin to cap the results since 2D CNN cannot capture temporal information and 3D CNN are to much resource demanding for real-time applications. Capsule Network, evolution of CNN, already proves its interesting benefits on small and low informational datasets like MNIST but yet its true potential has not emerged. In this paper we tackle the action recognition problem by proposing a new architecture combining Temporal Shift module over deep Capsule Network. Temporal Shift module permits us to insert temporal information over 2D Capsule Network with a zero computational cost to conserve the lightness of 2D capsules and their ability to connect spatial features. Our proposed approach outperforms or brings near state-of-the-art results on color and depth information on public datasets like First Person Hand Action and DHG 14/28 with a number of parameters 10 to 40 times less than existing approaches.

A Grid-Based Representation for Human Action Recognition

Soufiane Lamghari, Guillaume-Alexandre Bilodeau, Nicolas Saunier

Responsive image

Auto-TLDR; GRAR: Grid-based Representation for Action Recognition in Videos

Slides Poster Similar

Human action recognition (HAR) in videos is a fundamental research topic in computer vision. It consists mainly in understanding actions performed by humans based on a sequence of visual observations. In recent years, HAR have witnessed significant progress, especially with the emergence of deep learning models. However, most of existing approaches for action recognition rely on information that is not always relevant for the task, and are limited in the way they fuse temporal information. In this paper, we propose a novel method for human action recognition that encodes efficiently the most discriminative appearance information of an action with explicit attention on representative pose features, into a new compact grid representation. Our GRAR (Grid-based Representation for Action Recognition) method is tested on several benchmark datasets that demonstrate that our model can accurately recognize human actions, despite intra-class appearance variations and occlusion challenges.

Extracting Action Hierarchies from Action Labels and their Use in Deep Action Recognition

Konstadinos Bacharidis, Antonis Argyros

Responsive image

Auto-TLDR; Exploiting the Information Content of Language Label Associations for Human Action Recognition

Slides Poster Similar

Human activity recognition is a fundamental and challenging task in computer vision. Its solution can support multiple and diverse applications in areas including but not limited to smart homes, surveillance, daily living assistance, Human-Robot Collaboration (HRC), etc. In realistic conditions, the complexity of human activities ranges from simple coarse actions, such as siting or standing up, to more complex activities that consist of multiple actions with subtle variations in appearance and motion patterns. A large variety of existing datasets target specific action classes, with some of them being coarse and others being fine-grained. In all of them, a description of the action and its complexity is manifested in the action label sentence. As the action/activity complexity increases, so is the label sentence size and the amount of action-related semantic information contained in this description. In this paper, we propose an approach to exploit the information content of these action labels to formulate a coarse-to-fine action hierarchy based on linguistic label associations, and investigate the potential benefits and drawbacks. Moreover, in a series of quantitative and qualitative experiments, we show that the exploitation of this hierarchical organization of action classes in different levels of granularity improves the learning speed and overall performance of a range of baseline and mid-range deep architectures for human action recognition (HAR).

An Improved Bilinear Pooling Method for Image-Based Action Recognition

Wei Wu, Jiale Yu

Responsive image

Auto-TLDR; An improved bilinear pooling method for image-based action recognition

Slides Poster Similar

Action recognition in still images is a challenging task because of the complexity of human motions and the variance of background in the same action category. And some actions typically occur in fine-grained categories, with little visual differences between these categories. So extracting discriminative features or modeling various semantic parts is essential for image-based action recognition. Many methods apply expensive manual annotations to learn discriminative parts information for action recognition, which may severely discourage potential applications in real life. In recent years, bilinear pooling method has shown its effectiveness for image classification due to its learning distinctive features automatically. Inspired by this model, in this paper, an improved bilinear pooling method is proposed for avoiding the shortcomings of traditional bilinear pooling methods. The previous bilinear pooling approaches contain lots of noisy background or harmful feature information, which limit their application for action recognition. In our method, the attention mechanism is introduced into hierarchical bilinear pooling framework with mask aggregation for action recognition. The proposed model can generate the distinctive and ROI-aware feature information by combining multiple attention mask maps from the channel and spatial-wise attention features. To be more specific, our method makes the network to better pay attention to discriminative region of the vital objects in an image. We verify our model on the two challenging datasets: 1) Stanford 40 action dataset and 2) our action dataset that includes 60 categories. Experimental results demonstrate the effectiveness of our approach, which is superior to the traditional and state-of-the-art methods.

Attention As Activation

Yimian Dai, Stefan Oehmcke, Fabian Gieseke, Yiquan Wu, Kobus Barnard

Responsive image

Auto-TLDR; Attentional Activation Units for Convolutional Networks

Slides Similar

Activation functions and attention mechanisms are typically treated as having different purposes and have evolved differently. However, both concepts can be formulated as a non-linear gating function. Inspired by their similarity, we propose a novel type of activation units called attentional activation~(ATAC) units as a unification of activation functions and attention mechanisms. In particular, we propose a local channel attention module for the simultaneous non-linear activation and element-wise feature refinement, which locally aggregates point-wise cross-channel feature contexts. By replacing the well-known rectified linear units by such ATAC units in convolutional networks, we can construct fully attentional networks that perform significantly better with a modest number of additional parameters. We conducted detailed ablation studies on the ATAC units using several host networks with varying network depths to empirically verify the effectiveness and efficiency of the units. Furthermore, we compared the performance of the ATAC units against existing activation functions as well as other attention mechanisms on the CIFAR-10, CIFAR-100, and ImageNet datasets. Our experimental results show that networks constructed with the proposed ATAC units generally yield performance gains over their competitors given a comparable number of parameters.

ESResNet: Environmental Sound Classification Based on Visual Domain Models

Andrey Guzhov, Federico Raue, Jörn Hees, Andreas Dengel

Responsive image

Auto-TLDR; Environmental Sound Classification with Short-Time Fourier Transform Spectrograms

Slides Poster Similar

Environmental Sound Classification (ESC) is an active research area in the audio domain and has seen a lot of progress in the past years. However, many of the existing approaches achieve high accuracy by relying on domain-specific features and architectures, making it harder to benefit from advances in other fields (e.g., the image domain). Additionally, some of the past successes have been attributed to a discrepancy of how results are evaluated (i.e., on unofficial splits of the UrbanSound8K (US8K) dataset), distorting the overall progression of the field. The contribution of this paper is twofold. First, we present a model that is inherently compatible with mono and stereo sound inputs. Our model is based on simple log-power Short-Time Fourier Transform (STFT) spectrograms and combines them with several well-known approaches from the image domain (i.e., ResNet, Siamese-like networks and attention). We investigate the influence of cross-domain pre-training, architectural changes, and evaluate our model on standard datasets. We find that our model out-performs all previously known approaches in a fair comparison by achieving accuracies of 97.0 % (ESC-10), 91.5 % (ESC-50) and 84.2 % / 85.4 % (US8K mono / stereo). Second, we provide a comprehensive overview of the actual state of the field, by differentiating several previously reported results on the US8K dataset between official or unofficial splits. For better reproducibility, our code (including any re-implementations) is made available.

Hierarchical Multimodal Attention for Deep Video Summarization

Melissa Sanabria, Frederic Precioso, Thomas Menguy

Responsive image

Auto-TLDR; Automatic Summarization of Professional Soccer Matches Using Event-Stream Data and Multi- Instance Learning

Slides Poster Similar

The way people consume sports on TV has drastically evolved in the last years, particularly under the combined effects of the legalization of sport betting and the huge increase of sport analytics. Several companies are nowadays sending observers in the stadiums to collect live data of all the events happening on the field during the match. Those data contain meaningful information providing a very detailed description of all the actions occurring during the match to feed the coaches and staff, the fans, the viewers, and the gamblers. Exploiting all these data, sport broadcasters want to generate extra content such as match highlights, match summaries, players and teams analytics, etc., to appeal subscribers. This paper explores the problem of summarizing professional soccer matches as automatically as possible using both the aforementioned event-stream data collected from the field and the content broadcasted on TV. We have designed an architecture, introducing first (1) a Multiple Instance Learning method that takes into account the sequential dependency among events and then (2) a hierarchical multimodal attention layer that grasps the importance of each event in an action. We evaluate our approach on matches from two professional European soccer leagues, showing its capability to identify the best actions for automatic summarization by comparing with real summaries made by human operators.

SAT-Net: Self-Attention and Temporal Fusion for Facial Action Unit Detection

Zhihua Li, Zheng Zhang, Lijun Yin

Responsive image

Auto-TLDR; Temporal Fusion and Self-Attention Network for Facial Action Unit Detection

Slides Poster Similar

Research on facial action unit detection has shown remarkable performances by using deep spatial learning models in recent years, however, it is far from reaching its full capacity in learning due to the lack of use of temporal information of AUs across time. Since the AU occurrence in one frame is highly likely related to previous frames in a temporal sequence, exploring temporal correlation of AUs across frames becomes a key motivation of this work. In this paper, we propose a novel temporal fusion and AU-supervised self-attention network (a so-called SAT-Net) to address the AU detection problem. First of all, we input the deep features of a sequence into a convolutional LSTM network and fuse the previous temporal information into the feature map of the last frame, and continue to learn the AU occurrence. Second, considering the AU detection problem is a multi-label classification problem that individual label depends only on certain facial areas, we propose a new self-learned attention mask by focusing the detection of each AU on parts of facial areas through the learning of individual attention mask for each AU, thus increasing the AU independence without the loss of any spatial relations. Our extensive experiments show that the proposed framework achieves better results of AU detection over the state-of-the-arts on two benchmark databases (BP4D and DISFA).

Late Fusion of Bayesian and Convolutional Models for Action Recognition

Camille Maurice, Francisco Madrigal, Frederic Lerasle

Responsive image

Auto-TLDR; Fusion of Deep Neural Network and Bayesian-based Approach for Temporal Action Recognition

Slides Poster Similar

The activities we do in our daily-life are generally carried out as a succession of atomic actions, following a logical order. During a video sequence, actions usually follow a logical order. In this paper, we propose a hybrid approach resulting from the fusion of a deep learning neural network with a Bayesian-based approach. The latter models human-object interactions and transition between actions. The key idea is to combine both approaches in the final prediction. We validate our strategy in two public datasets: CAD-120 and Watch-n-Patch. We show that our fusion approach yields performance gains in accuracy of respectively +4\% and +6\% over a baseline approach. Temporal action recognition performances are clearly improved by the fusion, especially when classes are imbalanced.

You Ought to Look Around: Precise, Large Span Action Detection

Ge Pan, Zhang Han, Fan Yu, Yonghong Song, Yuanlin Zhang, Han Yuan

Responsive image

Auto-TLDR; YOLA: Local Feature Extraction for Action Localization with Variable receptive field

Slides Similar

For the action localization task, pre-defined action anchors are the cornerstone of mainstream techniques. State-of-the-art models mostly rely on a dense segmenting scheme, where anchors are sampled uniformly over the temporal domain with a predefined set of scales. However, it is not sufficient because action duration varies greatly. Therefore, it is necessary for the anchors or proposals to have a variable receptive field. In this paper, we propose a method called YOLA (You Ought to Look Around) which includes three parts: 1) a robust backbone SPN-I3D for extracting spatio-temporal features. In this part, we employ a stronger backbone I3D with SPN (Segment Pyramid Network) instead of C3D to obtain multi-scale features; 2) a simple but useful feature fusion module named LFE (Local Feature Extraction). Compared with the fully connected layer and global average pooling, our LFE model is more advantageous for network to fit and fuse features. 3) a new feature segment aligning method called TPGC (Two Pathway Graph Convolution), which allows one proposal to leverage semantic features of adjacent proposals to update its content and make sure the proposals have a variable receptive field. YOLA add only a small overhead to the baseline network, and is easy to train in an end-to-end manner, running at a speed of 1097 fps. YOLA achieves a mAP of 58.3%, outperforming all existing models including both RGB-based and two stream on THUMOS'14, and achieves competitive results on ActivityNet 1.3.

Towards Practical Compressed Video Action Recognition: A Temporal Enhanced Multi-Stream Network

Bing Li, Longteng Kong, Dongming Zhang, Xiuguo Bao, Di Huang, Yunhong Wang

Responsive image

Auto-TLDR; TEMSN: Temporal Enhanced Multi-Stream Network for Compressed Video Action Recognition

Slides Poster Similar

Current compressed video action recognition methods are mainly based on completely received compressed videos. However, in real transmission, the compressed video packets are usually disorderly received and lost due to network jitters or congestion. It is of great significance to recognize actions in early phases with limited packets, e.g. forecasting the potential risks from videos quickly. In this paper, we proposed a Temporal Enhanced Multi-Stream Network (TEMSN) for practical compressed video action recognition. First, we use three compressed modalities as complementary cues and build a multi-stream network to capture the rich information from compressed video packets. Second, we design a temporal enhanced module based on Encoder-Decoder structure applied on each stream to infer the missing packets, and generate more complete action dynamics. Thanks to the rich modalities and temporal enhancement, our approach is able to better modeling the action with limited compressed packets. Experiments on HMDB-51 and UCF-101 dataset validate its effectiveness and efficiency.

Temporal Attention-Augmented Graph Convolutional Network for Efficient Skeleton-Based Human Action Recognition

Negar Heidari, Alexandros Iosifidis

Responsive image

Auto-TLDR; Temporal Attention Module for Efficient Graph Convolutional Network-based Action Recognition

Slides Poster Similar

Graph convolutional networks (GCNs) have been very successful in modeling non-Euclidean data structures, like sequences of body skeletons forming actions modeled as spatio-temporal graphs. Most GCN-based action recognition methods use deep feed-forward networks with high computational complexity to process all skeletons in an action. This leads to a high number of floating point operations (ranging from 16G to 100G FLOPs) to process a single sample, making their adoption in restricted computation application scenarios infeasible. In this paper, we propose a temporal attention module (TAM) for increasing the efficiency in skeleton-based action recognition by selecting the most informative skeletons of an action at the early layers of the network. We incorporate the TAM in a light-weight GCN topology to further reduce the overall number of computations. Experimental results on two benchmark datasets show that the proposed method outperforms with a large margin the baseline GCN-based method while having 2.9 times less number of computations. Moreover, it performs on par with the state-of-the-art with up to 9.6 times less number of computations.

Vertex Feature Encoding and Hierarchical Temporal Modeling in a Spatio-Temporal Graph Convolutional Network for Action Recognition

Konstantinos Papadopoulos, Enjie Ghorbel, Djamila Aouada, Bjorn Ottersten

Responsive image

Auto-TLDR; Spatio-Temporal Graph Convolutional Network for Skeleton-Based Action Recognition

Slides Poster Similar

Spatio-temporal Graph Convolutional Networks (ST-GCNs) have shown great performance in the context of skeleton-based action recognition. Nevertheless, ST-GCNs use raw skeleton data as vertex features. Such features have low dimensionality and might not be optimal for action discrimination. Moreover, a single layer of temporal convolution is used to model short-term temporal dependencies but can be insufficient for capturing both long-term. In this paper, we extend the Spatio-Temporal Graph Convolutional Network for skeleton-based action recognition by introducing two novel modules, namely, the Graph Vertex Feature Encoder (GVFE) and the Dilated Hierarchical Temporal Convolutional Network (DH-TCN). On the one hand, the GVFE module learns appropriate vertex features for action recognition by encoding raw skeleton data into a new feature space. On the other hand, the DH-TCN module is capable of capturing both short-term and long-term temporal dependencies using a hierarchical dilated convolutional network. Experiments have been conducted on the challenging NTU RGB-D 60, NTU RGB-D 120 and Kinetics datasets. The obtained results show that our method competes with state-of-the-art approaches while using a smaller number of layers and parameters; thus reducing the required training time and memory.

Attention Pyramid Module for Scene Recognition

Zhinan Qiao, Xiaohui Yuan, Chengyuan Zhuang, Abolfazl Meyarian

Responsive image

Auto-TLDR; Attention Pyramid Module for Multi-Scale Scene Recognition

Slides Poster Similar

The unrestricted open vocabulary and diverse substances of scenery images bring significant challenges to scene recognition. However, most deep learning architectures and attention methods are developed on general-purpose datasets and omit the characteristics of scene data. In this paper, we exploit the attention pyramid module (APM) to tackle the predicament of scene recognition. Our method streamlines the multi-scale scene recognition pipeline, learns comprehensive scene features at various scales and locations, addresses the interdependency among scales, and further assists feature re-calibration as well as aggregation process. APM is extremely light-weighted and can be easily plugged into existing network architectures in a parameter-efficient manner. By simply integrating APM into ResNet-50, we obtain a 3.54\% boost in terms of top-1 accuracy on the benchmark scene dataset. Comprehensive experiments show that APM achieves better performance comparing with state-of-the-art attention methods using significant less computation budget. Code and pre-trained models will be made publicly available.

Extraction and Analysis of 3D Kinematic Parameters of Table Tennis Ball from a Single Camera

Jordan Calandre, Renaud Péteri, Laurent Mascarilla, Benoit Tremblais

Responsive image

Auto-TLDR; 3D Ball Trajectories Analysis using a Single Camera for Sport Gesture Analysis

Slides Poster Similar

Vision is the first indicator for coaches to assess the quality of a sport gesture. However, gesture analysis using computer vision is often restricted to laboratory experiments, far from the real conditions in which athletes train on a daily basis. In this perspective, we introduce 3D ball trajectories analysis using a single camera with very few acquisition constraints. A key point of the proposal is the estimation of the apparent ball size for obtaining ball to camera distance. For this purpose, a 2D CNN is trained using a generated dataset that enables a reliable ball size extraction, even in case of high motion blur. The final objective is not only to be able to determine ball trajectories, but most importantly to retrieve their relevant physical parameters. With a precise estimation of those trajectories, it is indeed possible to extract the ball tangential and rotation speed, related to the so-called Magnus effect. Validation experiments for characterizing table tennis strokes are presented on both a synthetic dataset and on real video sequences.

Vision-Based Multi-Modal Framework for Action Recognition

Djamila Romaissa Beddiar, Mourad Oussalah, Brahim Nini

Responsive image

Auto-TLDR; Multi-modal Framework for Human Activity Recognition Using RGB, Depth and Skeleton Data

Slides Poster Similar

Human activity recognition plays a central role in the development of intelligent systems for video surveillance, public security, health care and home monitoring, where detection and recognition of activities can improve the quality of life and security of humans. Typically, automated, intuitive and real-time systems are required to recognize human activities and identify accurately unusual behaviors in order to prevent dangerous situations. In this work, we explore the combination of three modalities (RGB, depth and skeleton data) to design a robust multi-modal framework for vision-based human activity recognition. Especially, spatial information, body shape/posture and temporal evolution of actions are highlighted using illustrative representations obtained from a combination of dynamic RGB images, dynamic depth images and skeleton data representations. Therefore, each video is represented with three images that summarize the ongoing action. Our framework takes advantage of transfer learning from pre trained models to extract significant features from these newly created images. Next, we fuse extracted features using Canonical Correlation Analysis and train a Long Short-Term Memory network to classify actions from visual descriptive images. Experimental results demonstrated the reliability of our feature-fusion framework that allows us to capture highly significant features and enables us to achieve the state-of-the-art performance on the public UTD-MHAD and NTU RGB+D datasets.

Wavelet Attention Embedding Networks for Video Super-Resolution

Young-Ju Choi, Young-Woon Lee, Byung-Gyu Kim

Responsive image

Auto-TLDR; Wavelet Attention Embedding Network for Video Super-Resolution

Slides Poster Similar

Recently, Video super-resolution (VSR) has become more crucial as the resolution of display has been grown. The majority of deep learning-based VSR methods combine the convolutional neural networks (CNN) with motion compensation or alignment module to estimate high-resolution (HR) frame from low-resolution (LR) frames. However, most of previous methods deal with the spatial features equally and may result in the misaligned temporal features by pixel-based motion compensation and alignment module. It can lead to the damaging effect on the accuracy of the estimated HR feature. In this paper, we propose a wavelet attention embedding network (WAEN), including wavelet embedding network (WENet) and attention embedding network (AENet), to fully exploit the spatio-temporal informative features. The WENet is operated as a spatial feature extractor of individual low and high-frequency information based on 2-D Haar discrete wavelet transform. The meaningful temporal feature is extracted in the AENet through utilizing the weighted attention map between frames. Experimental results demonstrate that the proposed method achieves superior performance compared with state-of-the-art methods.

Motion U-Net: Multi-Cue Encoder-Decoder Network for Motion Segmentation

Gani Rahmon, Filiz Bunyak, Kannappan Palaniappan

Responsive image

Auto-TLDR; Motion U-Net: A Deep Learning Framework for Robust Moving Object Detection under Challenging Conditions

Slides Poster Similar

Detection of moving objects is a critical first step in many computer vision applications. Several algorithms for motion and change detection were proposed. However, many of these approaches lack the ability to handle challenging real-world scenarios. Recently, deep learning approaches started to produce impressive solutions to computer vision tasks, particularly for detection and segmentation. Many existing deep learning networks proposed for moving object detection rely only on spatial appearance cues. In this paper, we propose a novel multi-cue and multi-stream network, Motion U-Net (MU-Net), which integrates motion, change, and appearance cues using a deep learning framework for robust moving object detection under challenging conditions. The proposed network consists of a two-stream encoder module followed by feature concatenation and a decoder module. Motion and change cues are computed through our tensor-based motion estimation and a multi-modal background subtraction modules. The proposed system was tested and evaluated on the change detection challenge datasets (CDnet-2014) and compared to state-of-the-art methods. On CDnet-2014 dataset, our approach reaches an average overall F-measure of 0.9852 and outperforms all current state-of-the-art methods. The network was also tested on the unseen SBI-2015 dataset and produced promising results.

Video Face Manipulation Detection through Ensemble of CNNs

Nicolo Bonettini, Edoardo Daniele Cannas, Sara Mandelli, Luca Bondi, Paolo Bestagini, Stefano Tubaro

Responsive image

Auto-TLDR; Face Manipulation Detection in Video Sequences Using Convolutional Neural Networks

Slides Similar

In the last few years, several techniques for facial manipulation in videos have been successfully developed and made available to the masses (i.e., FaceSwap, deepfake, etc.). These methods enable anyone to easily edit faces in video sequences with incredibly realistic results and a very little effort. Despite the usefulness of these tools in many fields, if used maliciously, they can have a significantly bad impact on society (e.g., fake news spreading, cyber bullying through fake revenge porn). The ability of objectively detecting whether a face has been manipulated in a video sequence is then a task of utmost importance. In this paper, we tackle the problem of face manipulation detection in video sequences targeting modern facial manipulation techniques. In particular, we study the ensembling of different trained Convolutional Neural Network (CNN) models. In the proposed solution, different models are obtained starting from a base network (i.e., EfficientNetB4) making use of two different concepts: (i) attention layers; (ii) siamese training. We show that combining these networks leads to promising face manipulation detection results on two publicly available datasets with more than 119000 videos.

Gabriella: An Online System for Real-Time Activity Detection in Untrimmed Security Videos

Mamshad Nayeem Rizve, Ugur Demir, Praveen Praveen Tirupattur, Aayush Jung Rana, Kevin Duarte, Ishan Rajendrakumar Dave, Yogesh Rawat, Mubarak Shah

Responsive image

Auto-TLDR; Gabriella: A Real-Time Online System for Activity Detection in Surveillance Videos

Slides Similar

Activity detection in surveillance videos is a difficult problem due to multiple factors such as large field of view, presence of multiple activities, varying scales and viewpoints, and its untrimmed nature. The existing research in activity detection is mainly focused on datasets, such as UCF-101, JHMDB, THUMOS, and AVA, which partially address these issues. The requirement of processing the surveillance videos in real-time makes this even more challenging. In this work we propose Gabriella, a real-time online system to perform activity detection on untrimmed surveillance videos. The proposed method consists of three stages: tubelet extraction, activity classification, and online tubelet merging. For tubelet extraction, we propose a localization network which takes a video clip as input and spatio-temporally detects potential foreground regions at multiple scales to generate action tubelets. We propose a novel Patch-Dice loss to handle large variations in actor size. Our online processing of videos at a clip level drastically reduces the computation time in detecting activities. The detected tubelets are assigned activity class scores by the classification network and merged together using our proposed Tubelet-Merge Action-Split (TMAS) algorithm to form the final action detections. The TMAS algorithm efficiently connects the tubelets in an online fashion to generate action detections which are robust against varying length activities. We perform our experiments on the VIRAT and MEVA (Multiview Extended Video with Activities) datasets and demonstrate the effectiveness of the proposed approach in terms of speed ($\sim$100 fps) and performance with state-of-the-art results. The code and models will be made publicly available.

Two-Level Attention-Based Fusion Learning for RGB-D Face Recognition

Hardik Uppal, Alireza Sepas-Moghaddam, Michael Greenspan, Ali Etemad

Responsive image

Auto-TLDR; Fused RGB-D Facial Recognition using Attention-Aware Feature Fusion

Slides Poster Similar

With recent advances in RGB-D sensing technologies as well as improvements in machine learning and fusion techniques, RGB-D facial recognition has become an active area of research. A novel attention aware method is proposed to fuse two image modalities, RGB and depth, for enhanced RGB-D facial recognition. The proposed method first extracts features from both modalities using a convolutional feature extractor. These features are then fused using a two layer attention mechanism. The first layer focuses on the fused feature maps generated by the feature extractor, exploiting the relationship between feature maps using LSTM recurrent learning. The second layer focuses on the spatial features of those maps using convolution. The training database is preprocessed and augmented through a set of geometric transformations, and the learning process is further aided using transfer learning from a pure 2D RGB image training process. Comparative evaluations demonstrate that the proposed method outperforms other state-of-the-art approaches, including both traditional and deep neural network-based methods, on the challenging CurtinFaces and IIIT-D RGB-D benchmark databases, achieving classification accuracies over 98.2% and 99.3% respectively. The proposed attention mechanism is also compared with other attention mechanisms, demonstrating more accurate results.

Efficient-Receptive Field Block with Group Spatial Attention Mechanism for Object Detection

Jiacheng Zhang, Zhicheng Zhao, Fei Su

Responsive image

Auto-TLDR; E-RFB: Efficient-Receptive Field Block for Deep Neural Network for Object Detection

Slides Poster Similar

Object detection has been paid rising attention in computer vision field. Convolutional Neural Networks (CNNs) extract high-level semantic features of images, which directly determine the performance of object detection. As a common solution, embedding integration modules into CNNs can enrich extracted features and thereby improve the performance. However, the instability and inconsistency of internal multiple branches exist in these modules. To address this problem, we propose a novel multibranch module called Efficient-Receptive Field Block (E-RFB), in which multiple levels of features are combined for network optimization. Specifically, by downsampling and increasing depth, the E-RFB provides sufficient RF. Second, in order to eliminate the inconsistency across different branches, a novel spatial attention mechanism, namely, Group Spatial Attention Module (GSAM) is proposed. The GSAM gradually narrows a feature map by channel grouping; thus it encodes the information between spatial and channel dimensions into the final attention heat map. Third, the proposed module can be easily joined in various CNNs to enhance feature representation as a plug-and-play component. With SSD-style detectors, our method halves the parameters of the original detection head and achieves high accuracy on the PASCAL VOC and MS COCO datasets. Moreover, the proposed method achieves superior performance compared with state-of-the-art methods based on similar framework.

Contextual Classification Using Self-Supervised Auxiliary Models for Deep Neural Networks

Sebastian Palacio, Philipp Engler, Jörn Hees, Andreas Dengel

Responsive image

Auto-TLDR; Self-Supervised Autogenous Learning for Deep Neural Networks

Slides Poster Similar

Classification problems solved with deep neural networks (DNNs) typically rely on a closed world paradigm, and optimize over a single objective (e.g., minimization of the cross- entropy loss). This setup dismisses all kinds of supporting signals that can be used to reinforce the existence or absence of particular patterns. The increasing need for models that are interpretable by design makes the inclusion of said contextual signals a crucial necessity. To this end, we introduce the notion of Self-Supervised Autogenous Learning (SSAL). A SSAL objective is realized through one or more additional targets that are derived from the original supervised classification task, following architectural principles found in multi-task learning. SSAL branches impose low-level priors into the optimization process (e.g., grouping). The ability of using SSAL branches during inference, allow models to converge faster, focusing on a richer set of class-relevant features. We equip state-of-the-art DNNs with SSAL objectives and report consistent improvements for all of them on CIFAR100 and Imagenet. We show that SSAL models outperform similar state-of-the-art methods focused on contextual loss functions, auxiliary branches and hierarchical priors.

Feature-Supervised Action Modality Transfer

Fida Mohammad Thoker, Cees Snoek

Responsive image

Auto-TLDR; Cross-Modal Action Recognition and Detection in Non-RGB Video Modalities by Learning from Large-Scale Labeled RGB Data

Slides Poster Similar

This paper strives for action recognition and detection in video modalities like RGB, depth maps or 3D-skeleton sequences when only limited modality-specific labeled examples are available. For the RGB, and derived optical-flow, modality many large-scale labeled datasets have been made available. They have become the de facto pre-training choice when recognizing or detecting new actions from RGB datasets that have limited amounts of labeled examples available. Unfortunately, large-scale labeled action datasets for other modalities are unavailable for pre-training. In this paper, our goal is to recognize actions from limited examples in non-RGB video modalities, by learning from large-scale labeled RGB data. To this end, we propose a two-step training process: (i) we extract action representation knowledge from an RGB-trained teacher network and adapt it to a non-RGB student network. (ii) we then fine-tune the transfer model with available labeled examples of the target modality. For the knowledge transfer we introduce feature-supervision strategies, which rely on unlabeled pairs of two modalities (the RGB and the target modality) to transfer feature level representations from the teacher to the the student network. Ablations and generalizations with two RGB source datasets and two non-RGB target datasets demonstrate that an optical-flow teacher provides better action transfer features than RGB for both depth maps and 3D-skeletons, even when evaluated on a different target domain, or for a different task. Compared to alternative cross-modal action transfer methods we show a good improvement in performance especially when labeled non-RGB examples to learn from are scarce.

Knowledge Distillation for Action Anticipation Via Label Smoothing

Guglielmo Camporese, Pasquale Coscia, Antonino Furnari, Giovanni Maria Farinella, Lamberto Ballan

Responsive image

Auto-TLDR; A Multi-Modal Framework for Action Anticipation using Long Short-Term Memory Networks

Slides Poster Similar

Human capability to anticipate near future from visual observations and non-verbal cues is essential for developing intelligent systems that need to interact with people. Several research areas, such as human-robot interaction (HRI), assisted living or autonomous driving need to foresee future events to avoid crashes or help people. Egocentric scenarios are classic examples where action anticipation is applied due to their numerous applications. Such challenging task demands to capture and model domain's hidden structure to reduce prediction uncertainty. Since multiple actions may equally occur in the future, we treat action anticipation as a multi-label problem with missing labels extending the concept of label smoothing. This idea resembles the knowledge distillation process since useful information is injected into the model during training. We implement a multi-modal framework based on long short-term memory (LSTM) networks to summarize past observations and make predictions at different time steps. We perform extensive experiments on EPIC-Kitchens and EGTEA Gaze+ datasets including more than 2500 and 100 action classes, respectively. The experiments show that label smoothing systematically improves performance of state-of-the-art models for action anticipation.

ReADS: A Rectified Attentional Double Supervised Network for Scene Text Recognition

Qi Song, Qianyi Jiang, Xiaolin Wei, Nan Li, Rui Zhang

Responsive image

Auto-TLDR; ReADS: Rectified Attentional Double Supervised Network for General Scene Text Recognition

Slides Poster Similar

In recent years, scene text recognition is always regarded as a sequence-to-sequence problem. Connectionist Temporal Classification (CTC) and Attentional sequence recognition (Attn) are two very prevailing approaches to tackle this problem while they may fail in some scenarios respectively. CTC concentrates more on every individual character but is weak in text semantic dependency modeling. Attn based methods have better context semantic modeling ability while tends to overfit on limited training data. In this paper, we elaborately design a Rectified Attentional Double Supervised Network (ReADS) for general scene text recognition. To overcome the weakness of CTC and Attn, both of them are applied in our method but with different modules in two supervised branches which can make a complementary to each other. Moreover, effective spatial and channel attention mechanisms are introduced to eliminate background noise and extract valid foreground information. Finally, a simple rectified network is implemented to rectify irregular text. The ReADS can be trained end-to-end and only word-level annotations are required. Extensive experiments on various benchmarks verify the effectiveness of ReADS which achieves state-of-the-art performance.

Activity Recognition Using First-Person-View Cameras Based on Sparse Optical Flows

Peng-Yuan Kao, Yan-Jing Lei, Chia-Hao Chang, Chu-Song Chen, Ming-Sui Lee, Yi-Ping Hung

Responsive image

Auto-TLDR; 3D Convolutional Neural Network for Activity Recognition with FPV Videos

Slides Poster Similar

First-person-view (FPV) cameras are finding wide use in daily life to record activities and sports. In this paper, we propose a succinct and robust 3D convolutional neural network (CNN) architecture accompanied with an ensemble-learning network for activity recognition with FPV videos. The proposed 3D CNN is trained on low-resolution (32x32) sparse optical flows using FPV video datasets consisting of daily activities. According to the experimental results, our network achieves an average accuracy of 90%.

Attention-Oriented Action Recognition for Real-Time Human-Robot Interaction

Ziyang Song, Ziyi Yin, Zejian Yuan, Chong Zhang, Wanchao Chi, Yonggen Ling, Shenghao Zhang

Responsive image

Auto-TLDR; Attention-Oriented Multi-Level Network for Action Recognition in Interaction Scenes

Slides Poster Similar

Despite the notable progress made in action recognition tasks, not much work has been done in action recognition specifically for human-robot interaction. In this paper, we deeply explore the characteristics of the action recognition task in interaction scenes and propose an attention-oriented multi-level network framework to meet the need for real-time interaction. Specifically, a Pre-Attention network is employed to roughly focus on the interactor in the scene at low resolution firstly and then perform fine-grained pose estimation at high resolution. The other compact CNN receives the extracted skeleton sequence as input for action recognition, utilizing attention-like mechanisms to capture local spatial-temporal patterns and global semantic information effectively. To evaluate our approach, we construct a new action dataset specially for the recognition task in interaction scenes. Experimental results on our dataset and high efficiency (112 fps at 640 x 480 RGBD) on the mobile computing platform (Nvidia Jetson AGX Xavier) demonstrate excellent applicability of our method on action recognition in real-time human-robot interaction.

Continuous Sign Language Recognition with Iterative Spatiotemporal Fine-Tuning

Kenessary Koishybay, Medet Mukushev, Anara Sandygulova

Responsive image

Auto-TLDR; A Deep Neural Network for Continuous Sign Language Recognition with Iterative Gloss Recognition

Slides Poster Similar

This paper aims to develop a deep neural network for Continuous Sign Language Recognition (CSLR) with iterative Gloss Recognition (GR) fine-tuning. CSLR has been a popular research field in the last years and iterative optimization methods are well established. This paper introduces our proposed architecture involving Spatiotemporal feature-extraction model to segment useful ``gloss-unit" features and BiLSTM with CTC as a sequence model. Spatiotemporal Feature Extractor is used for both image features extraction and sequence length reduction. To this end, we compare different architectures for feature extraction and sequence model. In addition, we iteratively fine-tune feature extractor on gloss-unit video segments with alignments from the end2end model. During the iterative training, we use novel alignment correction technique, which is based on minimum transformations of Levenshtein distance. All the experiments were conducted on the RWTH-PHOENIX-Weather-2014 dataset.