关键词: RGB-T无人机目标跟踪, Transformer, 跨模态特征交互, 特征协同重构, 跨模态位置线索融合

Abstract: RGB-T Unmanned Aerial Vehicle (UAV) object tracking enhances tracking robustness in complex environments by fusing comple-mentary information from visible (RGB) and thermal infrared (TIR) modalities. However, existing methods neglect the noise inter-ference caused by modality gaps, which weakens the effectiveness of cross-modal feature complementarity and degrades the power of feature representation, limiting the performance of RGB-T UAV trackers. To address this issue, a feature-cooperative reconstruc-tion-based tracker is proposed. The core of the proposed method is to develop a feature-cooperative reconstruction module, consist-ing of a cross-modal interaction encoder and a feature reconstruction decoder. Specifically, the cross-modal interaction encoder em-ploys an adaptive feature interaction strategy to extract critical complementary information from the auxiliary modality while effec-tively suppressing cross-modal noise interference. The feature reconstruction decoder then utilizes the query features from the encod-er to guide the reconstruction of features, preserving modality-specific information while incorporating cross-modal complementary details, thereby enhancing feature representation. Additionally, to enhance target localization accuracy in dynamic scenes, a cross-modal location cue fusion module is proposed to integrate search regions from different modalities, thereby providing more precise localization cues. Finally, extensive experimental evaluations on two RGB-T UAV object tracking benchmark datasets (i.e., VTUAV and HiAL) as well as the LasHeR dataset are conducted. The results demonstrate that the proposed method significantly outperforms existing methods. Concretely, the proposed method achieves the improvements of 9.9% in success rate and 9.0% in precision, re-spectively, compared to HMFT on the VTUAV dataset.

Key words: RGB-T Object Tracking, Transformer, Cross-modal Feature Interaction, Feature-Cooperative Reconstruction, Cross-modal Location Cue Fusion