Brain-Inspired Stepwise Patch Merging for Vision Transformers

TL;DR

This paper introduces Stepwise Patch Merging (SPM), a brain-inspired hierarchical approach for Vision Transformers that improves global and local feature integration, leading to enhanced performance in vision tasks.

Contribution

The paper proposes SPM, a novel patch merging method combining multi-scale aggregation and guided local enhancement, inspired by brain mechanisms, to improve ViT hierarchical architecture.

Findings

SPM improves accuracy on ImageNet-1K, COCO, and ADE20K datasets.

SPM enhances dense prediction tasks like object detection and segmentation.

Combining SPM with different backbones yields further performance gains.

Abstract

The hierarchical architecture has become a mainstream design paradigm for Vision Transformers (ViTs), with Patch Merging serving as the pivotal component that transforms a columnar architecture into a hierarchical one. Drawing inspiration from the brain's ability to integrate global and local information for comprehensive visual understanding, we propose Stepwise Patch Merging (SPM), which enhances the subsequent attention mechanism's ability to 'see' better. SPM consists of Multi-Scale Aggregation (MSA) and Guided Local Enhancement (GLE) striking a proper balance between long-range dependency modeling and local feature enhancement. Extensive experiments conducted on benchmark datasets, including ImageNet-1K, COCO, and ADE20K, demonstrate that SPM significantly improves the performance of various models, particularly in dense prediction tasks such as object detection and semantic segmentation. Meanwhile, experiments show that combining SPM with different backbones can further improve performance. The code has been released at https://github.com/Yonghao-Yu/StepwisePatchMerging.