Search

Video retrieval using natural language queries has attracted increasing interest due to its relevance in real-world applications, from intelligent access in private media galleries to web-scale video search. Learning the cross-similarity of video and text in a joint embedding space is the dominant approach. To do so, a contrastive loss is usually employed because it organizes the embedding space by putting similar items close and dissimilar items far. This framework leads to competitive recall rates, as they solely focus on the rank of the groundtruth items. Yet, assessing the quality of the ranking list is of utmost importance when considering intelligent retrieval systems, since multiple items may share similar semantics, hence a high relevance. Moreover, the aforementioned framework uses a fixed margin to separate similar and dissimilar items, treating all non-groundtruth items as equally irrelevant. In this paper we propose to use a variable margin: we argue that varying the margin used during training based on how much relevant an item is to a given query, i.e. a relevance-based margin, easily improves the quality of the ranking lists measured through nDCG and mAP. We demonstrate the advantages of our technique using different models on EPIC-Kitchens-100 and YouCook2. We show that even if we carefully tuned the fixed margin, our technique (which does not have the margin as a hyper-parameter) would still achieve better performance. Finally, extensive ablation studies and qualitative analysis support the robustness of our approach. Code will be released at \url{https://github.com/aranciokov/RelevanceMargin-ICMR22}., Comment: Accepted for presentation at International Conference on Multimedia Retrieval (ICMR '22)

Convolutional Neural Networks are the de facto models for image recognition. However 3D CNNs, the straight forward extension of 2D CNNs for video recognition, have not achieved the same success on standard action recognition benchmarks. One of the main reasons for this reduced performance of 3D CNNs is the increased computational complexity requiring large scale annotated datasets to train them in scale. 3D kernel factorization approaches have been proposed to reduce the complexity of 3D CNNs. Existing kernel factorization approaches follow hand-designed and hard-wired techniques. In this paper we propose Gate-Shift-Fuse (GSF), a novel spatio-temporal feature extraction module which controls interactions in spatio-temporal decomposition and learns to adaptively route features through time and combine them in a data dependent manner. GSF leverages grouped spatial gating to decompose input tensor and channel weighting to fuse the decomposed tensors. GSF can be inserted into existing 2D CNNs to convert them into an efficient and high performing spatio-temporal feature extractor, with negligible parameter and compute overhead. We perform an extensive analysis of GSF using two popular 2D CNN families and achieve state-of-the-art or competitive performance on five standard action recognition benchmarks., Comment: Accepted to TPAMI. arXiv admin note: text overlap with arXiv:1912.00381

This report presents the technical details of our submission to the EPIC-Kitchens-100 Action Recognition Challenge 2021. To participate in the challenge we deployed spatio-temporal feature extraction and aggregation models we have developed recently: GSF and XViT. GSF is an efficient spatio-temporal feature extracting module that can be plugged into 2D CNNs for video action recognition. XViT is a convolution free video feature extractor based on transformer architecture. We design an ensemble of GSF and XViT model families with different backbones and pretraining to generate the prediction scores. Our submission, visible on the public leaderboard, achieved a top-1 action recognition accuracy of 44.82%, using only RGB., Comment: Ranked third in the EPIC-Kitchens-100 Action Recognition Challenge @ CVPR 2021

This paper is on video recognition using Transformers. Very recent attempts in this area have demonstrated promising results in terms of recognition accuracy, yet they have been also shown to induce, in many cases, significant computational overheads due to the additional modelling of the temporal information. In this work, we propose a Video Transformer model the complexity of which scales linearly with the number of frames in the video sequence and hence induces no overhead compared to an image-based Transformer model. To achieve this, our model makes two approximations to the full space-time attention used in Video Transformers: (a) It restricts time attention to a local temporal window and capitalizes on the Transformer's depth to obtain full temporal coverage of the video sequence. (b) It uses efficient space-time mixing to attend jointly spatial and temporal locations without inducing any additional cost on top of a spatial-only attention model. We also show how to integrate 2 very lightweight mechanisms for global temporal-only attention which provide additional accuracy improvements at minimal computational cost. We demonstrate that our model produces very high recognition accuracy on the most popular video recognition datasets while at the same time being significantly more efficient than other Video Transformer models. Code will be made available., Comment: Updated results on SSv2

We present EgoACO, a deep neural architecture for video action recognition that learns to pool action-context-object descriptors from frame level features by leveraging the verb-noun structure of action labels in egocentric video datasets. The core component of EgoACO is class activation pooling (CAP), a differentiable pooling operation that combines ideas from bilinear pooling for fine-grained recognition and from feature learning for discriminative localization. CAP uses self-attention with a dictionary of learnable weights to pool from the most relevant feature regions. Through CAP, EgoACO learns to decode object and scene context descriptors from video frame features. For temporal modeling in EgoACO, we design a recurrent version of class activation pooling termed Long Short-Term Attention (LSTA). LSTA extends convolutional gated LSTM with built-in spatial attention and a re-designed output gate. Action, object and context descriptors are fused by a multi-head prediction that accounts for the inter-dependencies between noun-verb-action structured labels in egocentric video datasets. EgoACO features built-in visual explanations, helping learning and interpretation. Results on the two largest egocentric action recognition datasets currently available, EPIC-KITCHENS and EGTEA, show that by explicitly decoding action-context-object descriptors, EgoACO achieves state-of-the-art recognition performance., Comment: Accepted to TPAMI

In this report we describe the technical details of our submission to the EPIC-Kitchens Action Recognition 2020 Challenge. To participate in the challenge we deployed spatio-temporal feature extraction and aggregation models we have developed recently: Gate-Shift Module (GSM) [1] and EgoACO, an extension of Long Short-Term Attention (LSTA) [2]. We design an ensemble of GSM and EgoACO model families with different backbones and pre-training to generate the prediction scores. Our submission, visible on the public leaderboard with team name FBK-HUPBA, achieved a top-1 action recognition accuracy of 40.0% on S1 setting, and 25.71% on S2 setting, using only RGB., Comment: Ranked 3rd in the EPIC-Kitchens action recognition challenge @ CVPR 2020

Deep 3D CNNs for video action recognition are designed to learn powerful representations in the joint spatio-temporal feature space. In practice however, because of the large number of parameters and computations involved, they may under-perform in the lack of sufficiently large datasets for training them at scale. In this paper we introduce spatial gating in spatial-temporal decomposition of 3D kernels. We implement this concept with Gate-Shift Module (GSM). GSM is lightweight and turns a 2D-CNN into a highly efficient spatio-temporal feature extractor. With GSM plugged in, a 2D-CNN learns to adaptively route features through time and combine them, at almost no additional parameters and computational overhead. We perform an extensive evaluation of the proposed module to study its effectiveness in video action recognition, achieving state-of-the-art results on Something Something-V1 and Diving48 datasets, and obtaining competitive results on EPIC-Kitchens with far less model complexity., Comment: CVPR20 camera ready version. Code and models available at https://github.com/swathikirans/GSM

We review three recent deep learning based methods for action recognition and present a brief comparative analysis of the methods from a neurophyisiological point of view. We posit that there are some analogy between the three presented deep learning based methods and some of the existing hypotheses regarding the functioning of human brain., Comment: Presented as an extended abstract in the Mutual benefits of cognitive and computer vision (MBCCV) workshop, CVPR 2019

In this report we describe the technical details of our submission to the EPIC-Kitchens 2019 action recognition challenge. To participate in the challenge we have developed a number of CNN-LSTA [3] and HF-TSN [2] variants, and submitted predictions from an ensemble compiled out of these two model families. Our submission, visible on the public leaderboard with team name FBK-HUPBA, achieved a top-1 action recognition accuracy of 35.54% on S1 setting, and 20.25% on S2 setting., Comment: Ranked 3rd in the EPIC-Kitchens 2019 action recognition challenge, held as part of CVPR 2019

Most action recognition methods base on a) a late aggregation of frame level CNN features using average pooling, max pooling, or RNN, among others, or b) spatio-temporal aggregation via 3D convolutions. The first assume independence among frame features up to a certain level of abstraction and then perform higher-level aggregation, while the second extracts spatio-temporal features from grouped frames as early fusion. In this paper we explore the space in between these two, by letting adjacent feature branches interact as they develop into the higher level representation. The interaction happens between feature differencing and averaging at each level of the hierarchy, and it has convolutional structure that learns to select the appropriate mode locally in contrast to previous works that impose one of the modes globally (e.g. feature differencing) as a design choice. We further constrain this interaction to be conservative, e.g. a local feature subtraction in one branch is compensated by the addition on another, such that the total feature flow is preserved. We evaluate the performance of our proposal on a number of existing models, i.e. TSN, TRN and ECO, to show its flexibility and effectiveness in improving action recognition performance.

Egocentric activity recognition is one of the most challenging tasks in video analysis. It requires a fine-grained discrimination of small objects and their manipulation. While some methods base on strong supervision and attention mechanisms, they are either annotation consuming or do not take spatio-temporal patterns into account. In this paper we propose LSTA as a mechanism to focus on features from spatial relevant parts while attention is being tracked smoothly across the video sequence. We demonstrate the effectiveness of LSTA on egocentric activity recognition with an end-to-end trainable two-stream architecture, achieving state of the art performance on four standard benchmarks., Comment: Accepted to CVPR 2019

Most recent approaches for action recognition from video leverage deep architectures to encode the video clip into a fixed length representation vector that is then used for classification. For this to be successful, the network must be capable of suppressing irrelevant scene background and extract the representation from the most discriminative part of the video. Our contribution builds on the observation that spatio-temporal patterns characterizing actions in videos are highly correlated with objects and their location in the video. We propose Top-down Attention Action VLAD (TA-VLAD), a deep recurrent architecture with built-in spatial attention that performs temporally aggregated VLAD encoding for action recognition from videos. We adopt a top-down approach of attention, by using class specific activation maps obtained from a deep CNN pre-trained for image classification, to weight appearance features before encoding them into a fixed-length video descriptor using Gated Recurrent Units. Our method achieves state of the art recognition accuracy on HMDB51 and UCF101 benchmarks., Comment: Accepted to the 17th International Conference of the Italian Association for Artificial Intelligence

In this paper we propose an end-to-end trainable deep neural network model for egocentric activity recognition. Our model is built on the observation that egocentric activities are highly characterized by the objects and their locations in the video. Based on this, we develop a spatial attention mechanism that enables the network to attend to regions containing objects that are correlated with the activity under consideration. We learn highly specialized attention maps for each frame using class-specific activations from a CNN pre-trained for generic image recognition, and use them for spatio-temporal encoding of the video with a convolutional LSTM. Our model is trained in a weakly supervised setting using raw video-level activity-class labels. Nonetheless, on standard egocentric activity benchmarks our model surpasses by up to +6% points recognition accuracy the currently best performing method that leverages hand segmentation and object location strong supervision for training. We visually analyze attention maps generated by the network, revealing that the network successfully identifies the relevant objects present in the video frames which may explain the strong recognition performance. We also discuss an extensive ablation analysis regarding the design choices., Comment: Accepted to BMVC 2018

Developing a technique for the automatic analysis of surveillance videos in order to identify the presence of violence is of broad interest. In this work, we propose a deep neural network for the purpose of recognizing violent videos. A convolutional neural network is used to extract frame level features from a video. The frame level features are then aggregated using a variant of the long short term memory that uses convolutional gates. The convolutional neural network along with the convolutional long short term memory is capable of capturing localized spatio-temporal features which enables the analysis of local motion taking place in the video. We also propose to use adjacent frame differences as the input to the model thereby forcing it to encode the changes occurring in the video. The performance of the proposed feature extraction pipeline is evaluated on three standard benchmark datasets in terms of recognition accuracy. Comparison of the results obtained with the state of the art techniques revealed the promising capability of the proposed method in recognizing violent videos., Comment: Accepted in International Conference on Advanced Video and Signal based Surveillance(AVSS 2017)

In this paper, we present a novel deep learning based approach for addressing the problem of interaction recognition from a first person perspective. The proposed approach uses a pair of convolutional neural networks, whose parameters are shared, for extracting frame level features from successive frames of the video. The frame level features are then aggregated using a convolutional long short-term memory. The hidden state of the convolutional long short-term memory, after all the input video frames are processed, is used for classification in to the respective categories. The two branches of the convolutional neural network perform feature encoding on a short time interval whereas the convolutional long short term memory encodes the changes on a longer temporal duration. In our network the spatio-temporal structure of the input is preserved till the very final processing stage. Experimental results show that our method outperforms the state of the art on most recent first person interactions datasets that involve complex ego-motion. In particular, on UTKinect-FirstPerson it competes with methods that use depth image and skeletal joints information along with RGB images, while it surpasses all previous methods that use only RGB images by more than 20% in recognition accuracy., Comment: Accepted on the second International Workshop on Egocentric Perception, Interaction and Computing(EPIC) at International Conference on Computer Vision(ICCV-17)