Learning Diverse Fine-Grained Features for Thermal Infrared Tracking

21 Pages Posted: 25 Apr 2023

See all articles by Chao Yang

Chao Yang

affiliation not provided to SSRN

Qiao Liu

affiliation not provided to SSRN

Gaojun Li

affiliation not provided to SSRN

Honghu Pan

affiliation not provided to SSRN

Zhenyu He

affiliation not provided to SSRN

Abstract

Existing feature models used in thermal infrared (TIR) tracking struggle to get strong discriminative feature of TIR objects, because TIR image has few details and low contrast. This characteristic makes the existing TIR tracking methods easy to drift to similar distractors. To tackle this problem, we introduce a novel diverse fine-grained feature network for TIR tracking. Our proposed method emphasizes extracting fine-grained features from multiple local regions of the target to improve its ability in discriminating against distractors. Specifically, our feature model consists of a specific-designed fine-grained feature network architecture and an auxiliary diverse loss function. Firstly, the fine-grained feature network learns to explore the subtle clues of the infrared target through a mask suppression mechanism. This mechanism can force the network to learn subtle cues of the target. Secondly, to guarantee the learned fine-grained features are various, we propose a diversity loss to force all fine-grained features are unique. These two modules help the feature model learn diverse fine-grained features from two complementary aspects. To prove that the method we proposed is effective, we evaluate it on four benchmarks. The relevant experimental results proves that our proposed method achieves the best performance compared to the state-of-the-art methods.

Keywords: Fine-grained features, Convolutional neural network, Transformer tracking, Thermal infrared tracking

Suggested Citation

Yang, Chao and Liu, Qiao and Li, Gaojun and Pan, Honghu and He, Zhenyu, Learning Diverse Fine-Grained Features for Thermal Infrared Tracking. Available at SSRN: https://ssrn.com/abstract=4429466 or http://dx.doi.org/10.2139/ssrn.4429466

Chao Yang

affiliation not provided to SSRN ( email )

No Address Available

Qiao Liu

affiliation not provided to SSRN ( email )

No Address Available

Gaojun Li

affiliation not provided to SSRN ( email )

No Address Available

Honghu Pan

affiliation not provided to SSRN ( email )

No Address Available

Zhenyu He (Contact Author)

affiliation not provided to SSRN ( email )

No Address Available

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
33
Abstract Views
97
PlumX Metrics