Abstract:Although the red, green, and blue (RGB) image has a high spatial resolution, it only depicts color intensities in RGB channels, which easily leads to the failure of the tracker based on RGB modality in some challenging scenarios, for example, when the color of the object and background is similar. The hyperspectral image with rich spectral information is more robust in these difficult situations, so it is essential to explore how to effectively apply hyperspectral features to supplement RGB information in object tracking. However, there is no fusion tracking algorithm based on hyperspectral and RGB data. Based on this, we propose a novel fusion tracking framework of hyperspectral and RGB in this article, termed as transformer-based fusion tracking network (TFTN), to enhance the performance of object tracking. Within the framework, we construct a dual-branch structure based on the Siamese network to obtain the modality-specific representations of different modality images. Besides, the framework is generic, which is suitable for the Siamese series of tracking algorithms. In addition, we design a Siamese 3-D convolutional neural network as the specific branch of hyperspectral modality for synchronous extraction of the spatial and spectral features of hyperspectral data, to give full play to the role of hyperspectral data in improving network tracking performance. Particularly, inspired by the structure of Transformer, we design a transformer-based fusion module to capture the potential interaction of intramodality and intermodality features of different modalities. This is the first work that combines the information of hyperspectral and RGB modalities to improve tracking performance. At the same time, it is also the first time that employs the self-attention module of Transformer to combine the information of different modalities for multimodality fusion tracking. Experimental results on the dataset composed of hyperspectral and RGB image sequences show that the proposed TFTN tracker is superior to the state-of-the-art trackers, demonstrating the effectiveness of this method.

Visible–Infrared Dual-Sensor Tracking Based on Transformer via Progressive Feature Enhancement and Fusion

Learning a Multimodal Feature Transformer for RGBT Tracking

Transformer-Based Dual-Modal Visual Target Tracking Using Visible Light and Thermal Infrared.

Dynamic Feature-Memory Transformer Network for RGBT Tracking

Highly Compact Adaptive Network Based on Transformer for RGBT Tracking

Unidirectional Cross-Modal Fusion for RGB-T Tracking

Visible and Infrared Object Tracking Based on Multimodal Hierarchical Relationship Modeling

Transformer-based RGB-T Tracking with Channel and Spatial Feature Fusion

Siamese transformer RGBT tracking

Real-Time RGBT Target Tracking Based on Attention Mechanism

RGB-T Tracking Based on Mixed Attention

Unified Single-Stage Transformer Network for Efficient RGB-T Tracking

Learning Modality Feature Fusion Via Transformer for RGBT-tracking

SiamCAF: Complementary Attention Fusion-Based Siamese Network for RGBT Tracking

TFTN: A Transformer-Based Fusion Tracking Framework of Hyperspectral and RGB

Efficient Transformer-based Single-stage RGB-T Tracking with Larger Search Region and Acceleration Mechanism

SiamTDR: Time-Efficient RGBT Tracking Via Disentangled Representations

Attention-Based Transformer Thermal Infrared Tracker

RGBT Image Fusion Tracking via Sparse Trifurcate Transformer Aggregation Network

Thermal Infrared Tracking Method Based on Efficient Global Information Perception

Multi-scale Feature Extraction and Fusion with Attention Interaction for RGB-T Tracking