End-to-End Object Detection with Adaptive Clustering Transformer

TL;DR

This paper introduces Adaptive Clustering Transformer (ACT), a novel method that reduces the computational cost of transformer-based object detection by adaptively clustering features, maintaining accuracy while lowering FLOPs.

Contribution

The paper proposes ACT, a new transformer variant that replaces standard self-attention with an efficient clustering-based approach, enabling faster training and inference for high-resolution inputs.

Findings

ACT reduces quadratic complexity to linear with respect to the number of prototypes.

ACT maintains competitive accuracy while significantly lowering FLOPs.

The method is a drop-in replacement for existing self-attention modules.

Abstract

End-to-end Object Detection with Transformer (DETR)proposes to perform object detection with Transformer and achieve comparable performance with two-stage object detection like Faster-RCNN. However, DETR needs huge computational resources for training and inference due to the high-resolution spatial input. In this paper, a novel variant of transformer named Adaptive Clustering Transformer(ACT) has been proposed to reduce the computation cost for high-resolution input. ACT cluster the query features adaptively using Locality Sensitive Hashing (LSH) and ap-proximate the query-key interaction using the prototype-key interaction. ACT can reduce the quadratic O(N2) complexity inside self-attention into O(NK) where K is the number of prototypes in each layer. ACT can be a drop-in module replacing the original self-attention module without any training. ACT achieves a good balance between accuracy and computation cost (FLOPs). The code is available as supplementary for the ease of experiment replication and verification. Code is released at \url{https://github.com/gaopengcuhk/SMCA-DETR/}