Related references
Note: Only part of the references are listed.
Article
Computer Science, Artificial Intelligence
Tao Zhao et al.
Summary: Weakly supervised temporal action localization is a widely studied topic in recent years, with existing methods categorized into pre-classification and post-classification pipelines. In this study, a unified framework is proposed to simultaneously learn these two pipelines using two parallel network streams and a shared classifier. The Equivalent Classification Mapping (ECM) mechanism is introduced to achieve accurate action localization results. The proposed framework also incorporates a weight-transition module and an equivalent training strategy to thoroughly mine the equivalence mechanism. Comprehensive experiments on three benchmarks demonstrate the effectiveness of ECM.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Zehua Sun et al.
Summary: Human Action Recognition (HAR) aims to understand human behavior and assign labels to actions. Various data modalities, such as RGB, skeleton, depth, infrared, point cloud, event stream, audio, acceleration, radar, and WiFi signal, can be used to represent human actions. Many studies have investigated different approaches for HAR using these modalities. This article presents a comprehensive survey of recent progress in deep learning methods for HAR based on input data modality, covering single and multiple modalities and fusion-based and co-learning-based frameworks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Engineering, Electrical & Electronic
Binglu Wang et al.
Summary: Modeling cross-video relationship at the sub-action granularity is proposed in this work for weakly supervised temporal action localization. By representing video features through a group of sub-actions (sub-action family) shared among all videos in the dataset, complicated sampling strategies can be eliminated. Additional top-down branch and consistency loss are introduced to learn feature vectors and guide the learning process. Experimental results on benchmark datasets demonstrate the high performance of the proposed method.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Engineering, Electrical & Electronic
Junyu Gao et al.
Summary: This paper proposes an alternative approach to video moment retrieval that does not require textual annotations of videos. It leverages existing visual concept detectors and a pre-trained image-sentence embedding space. By utilizing a video-conditioned sentence generator, a GNN-based relation-aware moment localizer, and a pre-trained image-sentence embedding space, the proposed method achieves effective video moment retrieval.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Le Yang et al.
Summary: This study proposes a novel method called BackTAL, which converts action-click supervision to background-click supervision, and trains a stronger action localizer on background video frames. BackTAL implements two-fold modeling on the background frames and dynamically attends to informative neighbors during temporal convolution. Extensive experiments demonstrate the high performance of BackTAL and the rationality of the proposed background-click supervision.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Theory & Methods
Pengzhen Ren et al.
Summary: Researchers have shown relatively lower interest in active learning compared to deep learning, but with the increasing demand for large-scale high-quality annotated datasets, active learning is receiving more attention. This article provides a comprehensive survey on deep active learning, including a formal classification method, an overview of existing work, and an analysis of developments from an application perspective.
ACM COMPUTING SURVEYS
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Sauradip Nag et al.
Summary: Existing temporal action detection methods heavily rely on annotated training data, which is expensive to collect. Semi-supervised temporal action detection (SS-TAD) addresses this issue by utilizing unlabeled videos. However, SS-TAD is a challenging and under-studied problem. In this work, we propose a novel SS-TAD model called SPOT, which eliminates the dependence between localization and classification through parallel architecture and introduces interaction mechanism and self-supervised pre-training. Our experiments demonstrate that SPOT outperforms state-of-the-art alternatives by a large margin.
COMPUTER VISION - ECCV 2022, PT III
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Xinpeng Ding et al.
Summary: The paper introduces a new segment-level supervision setting and proposes a local segment loss and propagation loss to address the temporal action localization problem.
PATTERN RECOGNITION AND COMPUTER VISION, PT I
(2021)
Article
Computer Science, Artificial Intelligence
Guozhang Li et al.
Summary: A novel supervision method named multi-hierarchical category supervision (MHCS) is introduced in this paper to encourage the model to pay more attention to common sub-actions rather than only discriminative ones in Weakly Supervised Temporal Action Localization (WTAL). By constructing super-classes through hierarchical clustering, the MHCS method improves the performance of WTAL models by finding more sub-actions shared among action categories.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Xinpeng Ding et al.
Summary: In this paper, a method named K-farthest crossover is proposed for semi-supervised learning in TAL, which helps construct perturbations based on video features. The method adds perturbations to features along the temporal axis and adopts CR to encourage the model to retain the observation of similarity and difference between features.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Proceedings Paper
Acoustics
Cece Jin et al.
2020 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING
(2020)
Article
Computer Science, Information Systems
Hao Song et al.
IEEE TRANSACTIONS ON MULTIMEDIA
(2019)
Proceedings Paper
Computer Science, Artificial Intelligence
Hamed H. Aghdam et al.
2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019)
(2019)
Proceedings Paper
Computer Science, Artificial Intelligence
Jingwei Ji et al.
2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019)
(2019)
Proceedings Paper
Computer Science, Software Engineering
Yupan Huang et al.
2019 IEEE INTERNATIONAL CONFERENCE ON MULTIMEDIA AND EXPO (ICME)
(2019)
Article
Engineering, Electrical & Electronic
Keze Wang et al.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Jiyang Gao et al.
2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Joao Carreira et al.
30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017)
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Navaneeth Bodla et al.
2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2017)
Article
Education & Educational Research
M Prince
JOURNAL OF ENGINEERING EDUCATION
(2004)