MViTv2: Improved Multiscale Vision Transformers for Classification and Detection

TL;DR

MViTv2 introduces an improved multiscale vision transformer architecture with decomposed relative positional embeddings and residual pooling, achieving state-of-the-art results across image classification, object detection, and video recognition tasks.

Contribution

The paper presents MViTv2, a unified multiscale vision transformer architecture with novel enhancements that outperform previous models in multiple visual recognition domains.

Findings

88.8% accuracy on ImageNet classification

58.7 boxAP on COCO detection

86.1% accuracy on Kinetics-400 video recognition

Abstract

In this paper, we study Multiscale Vision Transformers (MViTv2) as a unified architecture for image and video classification, as well as object detection. We present an improved version of MViT that incorporates decomposed relative positional embeddings and residual pooling connections. We instantiate this architecture in five sizes and evaluate it for ImageNet classification, COCO detection and Kinetics video recognition where it outperforms prior work. We further compare MViTv2s' pooling attention to window attention mechanisms where it outperforms the latter in accuracy/compute. Without bells-and-whistles, MViTv2 has state-of-the-art performance in 3 domains: 88.8% accuracy on ImageNet classification, 58.7 boxAP on COCO object detection as well as 86.1% on Kinetics-400 video classification. Code and models are available at https://github.com/facebookresearch/mvit.