Learning Dual-Fused Modality-Aware Representations for RGBD Tracking

Gao，Shang1; Yang，Jinyu1,2; Li，Zhe1; Zheng，Feng1; Leonardis，Aleš2; Song，Jingkuan3

10.1007/978-3-031-25085-9_27

2023

0302-9743

1611-3349

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

13808 LNCS

478-494

With the development of depth sensors in recent years, RGBD object tracking has received significant attention. Compared with the traditional RGB object tracking, the addition of the depth modality can effectively solve the target and background interference. However, some existing RGBD trackers use the two modalities separately and thus some particularly useful shared information between them is ignored. On the other hand, some methods attempt to fuse the two modalities by treating them equally, resulting in the missing of modality-specific features. To tackle these limitations, we propose a novel Dual-fused Modality-aware Tracker (termed DMTracker) which aims to learn informative and discriminative representations of the target objects for robust RGBD tracking. The first fusion module focuses on extracting the shared information between modalities based on cross-modal attention. The second aims at integrating the RGB-specific and depth-specific information to enhance the fused features. By fusing both the modality-shared and modality-specific information in a modality-aware scheme, our DMTracker can learn discriminative representations in complex tracking scenes. Experiments show that our proposed tracker achieves very promising results on challenging RGBD benchmarks. Code is available at https://github.com/ShangGaoG/DMTracker.

Multi-modal learning Object tracking RGBD tracking

第一 ; 通讯

英语

EI

20231413841789

Computer vision

Computer Applications:723.5 ; Vision:741.2

2-s2.0-85151390839

Scopus

1.Department of Computer Science and Engineering,Southern University of Science and Technology,Shenzhen,China
2.University of Birmingham,Birmingham,United Kingdom
3.University of Electronic Science and Technology of China,Chengdu,China

Gao，Shang,Yang，Jinyu,Li，Zhe,et al. Learning Dual-Fused Modality-Aware Representations for RGBD Tracking[C],2023:478-494.