SiamDCFF: Dynamic Cascade Feature Fusion for Vision Tracking

TL;DR

This paper introduces SiamDCFF, a new tracking method that improves feature fusion in Siamese networks by enhancing global dependencies.

Contribution

The novel dynamic cascade feature fusion module improves global dependency modeling in Siamese network-based trackers.

Findings

The DCFF module significantly enhances the performance of fully convolutional trackers.

SiamDCFF outperforms the baseline model on public tracking datasets.

Global dependencies in feature fusion lead to more robust tracking results.

Abstract

Establishing an accurate and robust feature fusion mechanism is key to enhancing the tracking performance of single-object trackers based on a Siamese network. However, the output features of the depth-wise cross-correlation feature fusion module in fully convolutional trackers based on Siamese networks cannot establish global dependencies on the feature maps of a search area. This paper proposes a dynamic cascade feature fusion (DCFF) module by introducing a local feature guidance (LFG) module and dynamic attention modules (DAMs) after the depth-wise cross-correlation module to enhance the global dependency modeling capability during the feature fusion process. In this paper, a set of verification experiments is designed to investigate whether establishing global dependencies for the features output by the depth-wise cross-correlation operation can significantly improve the performance of fully convolutional trackers based on a Siamese network, providing experimental support for rational design of the structure of a dynamic cascade feature fusion module. Secondly, we integrate the dynamic cascade feature fusion module into the tracking framework based on a Siamese network, propose SiamDCFF, and evaluate it using public datasets. Compared with the baseline model, SiamDCFF demonstrated significant improvements.

Figures