检索结果-南通市图书馆

Weakly supervised temporal action localization with proxy metric modeling

Frontiers of Computer Science 2023年第2期17卷 63-72页

作者： Hongsheng XU Zihan CHEN Yu ZHANG Xin GENG Siya MI Zhihong YANG NARI Group Corporation(State Grid Electric Power Research Institute) Nanjing 211106China School of Computer Science and Engineering and the Key Lab of Computer Network and Information Integration(Ministry of Education)Southeast UniversityNanjing 211189China School of Cyber Science and Engineering Southeast UniversityNanjing 211189China Purple Mountain Laboratories Nanjing 211111China

temporal localization is crucial for action video *** the manual annotations are expensive and time-consuming in videos,temporal localization with weak video-level labels is challenging but *** this paper,we propose a weakly-supervised temporal action localization approach in untrimmed *** settle this issue,we train the model based on the proxies of each action *** proxies are used to measure the distances between action segments and different original action *** use a proxy-based metric to cluster the same actions together and separate actions from *** with state-of-the-art methods,our method achieved competitive results on the THUMOS14 and ActivityNet1.2 datasets.

关键词： temporal action localization weakly supervised videos proxy metric

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Enhancing action discrimination via category-specific frame clustering for weakly-supervised temporal action localization

引用

Frontiers of Information Technology & Electronic Engineering 2024年第6期25卷 809-823页

作者： Huifen XIA Yongzhao ZHAN Honglin LIU Xiaopeng REN School of Computer Science and Communication Engineering Jiangsu UniversityZhenjiang 212013China Jiangsu Engineering Research Center of Big Data Ubiquitous Perception and Intelligent Agricultural Applications Zhenjiang 212013China Changzhou Vocational Institute of Mechatronic Technology Changzhou 213164China

temporal action localization (TAL) is a task of detecting the start and end timestamps of action instances and classifying them in an untrimmed video. As the number of action categories per video increases, existing weakly-supervised TAL (W-TAL) methods with only video-level labels cannot provide sufficient supervision. Single-frame supervision has attracted the interest of researchers. Existing paradigms model single-frame annotations from the perspective of video snippet sequences, neglect action discrimination of annotated frames, and do not pay sufficient attention to their correlations in the same category. Considering a category, the annotated frames exhibit distinctive appearance characteristics or clear action ***, a novel method to enhance action discrimination via category-specific frame clustering for W-TAL is proposed. Specifically,the K-means clustering algorithm is employed to aggregate the annotated discriminative frames of the same category, which are regarded as exemplars to exhibit the characteristics of the action category. Then, the class activation scores are obtained by calculating the similarities between a frame and exemplars of various categories. Category-specific representation modeling can provide complimentary guidance to snippet sequence modeling in the mainline. As a result, a convex combination fusion mechanism is presented for annotated frames and snippet sequences to enhance the consistency properties of action discrimination,which can generate a robust class activation sequence for precise action classification and localization. Due to the supplementary guidance of action discriminative enhancement for video snippet sequences, our method outperforms existing single-frame annotation based methods. Experiments conducted on three datasets (THUMOS14, GTEA, and BEOID) show that our method achieves high localization performance compared with state-of-the-art methods.

关键词： Weakly supervised temporal action localization Single-frame annotation Category-specific action discrimination

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Weakly-Supervised temporal action localization by Background Suppression

Weakly-Supervised Temporal Action Localization by Background...

引用

第40届中国控制会议

作者： Mengxue Liu Xiangjun Gao Fangzhen Ge Huaiyu Liu Wenjing Li Huaibei Normal University The school of computer science and technology

We propose a novel method of background suppression to solve the issue that background regions are recognized as actions in weakly-supervised temporal action localization. The general attention-based action localization methods tend to use the attention module to generate segment-level attention weights. But there is little difference between the attentions from the background segments similar to the target actions and the attentions of the action segments, which causes the result that many background segments related to the target actions are still recognized as actions. To address this issue, a weakly-supervised temporal action localization network by background suppression（BS-WTAL） is designed. It introduces a filtering module for suppressing the background features and encouraging the action features, a classification module for identifying action categories and a generative attention module for segment-wise representation modeling. This enables BS-WTAL to suppress background to improve localization performance. Furthermore, we conduct ablation studies from different perspectives. Extensive experiments were performed on two datasets – THUMOS14 and Activity Net1.2. Our approach shows better performance on these two datasets, even comparable with state-of-the-art fully-supervised methods.

关键词： Weak Supervision temporal action localization Filtering Module Background Suppression

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

欢迎您,

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

请选择保存的检索档案：

请选择收藏分类：

通借通还

欢迎您,

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：