HARDer-Net : Hardness-Guided Discrimination Network for 3D Early Activity Prediction

Li, Tianjiao and Luo, Yang and Zhang, Wei and Duan, Lingyu and Liu, Jun (2024) HARDer-Net : Hardness-Guided Discrimination Network for 3D Early Activity Prediction. IEEE Transactions on Circuits and Systems for Video Technology. ISSN 1051-8215

[thumbnail of HARDer_Net_csvt_final_version 16.10.05]
Text (HARDer_Net_csvt_final_version 16.10.05)
HARDer_Net_csvt_final_version_16.10.05.pdf - Accepted Version
Available under License Creative Commons Attribution.

Download (3MB)

Abstract

To predict the class label from a partially observable activity sequence can be quite challenging due to the high degree of similarity existing in early segments of different activities. In this paper, an innovative HARDness-Guided Discrimination Network (HARDer-Net) is proposed to evaluate the relationship between similar activity pairs that are extremely hard to discriminate. To train our HARDer-Net, an innovative adversarial learning scheme has been designed, providing our network with the strength to extract subtle discrimination information for the prediction of 3D early activities. Moreover, to enhance the adversarial learning scheme efficacy of our model for 3D early action prediction, we construct a Hardness-Guided bank that dynamically records the hard similar samples and conducts reward-guided selections of these recorded hard samples using a deep reinforcement learning scheme. The proposed method significantly enhances the capability of the model to discern fine-grained differences in early activity sequences. Several widely-used activity datasets are used to evaluate our proposed HARDer-Net, and we achieve state-of-the-art performance across all the evaluated datasets.

Item Type:
Journal Article
Journal or Publication Title:
IEEE Transactions on Circuits and Systems for Video Technology
Uncontrolled Keywords:
/dk/atira/pure/subjectarea/asjc/2200/2214
Subjects:
?? media technologyelectrical and electronic engineering ??
ID Code:
224205
Deposited By:
Deposited On:
01 Oct 2024 11:10
Refereed?:
Yes
Published?:
Published
Last Modified:
01 Oct 2024 11:10