ResNeSt: Split-Attention Networks

04/19/2020
by   Hang Zhang, et al.
16

While image classification models have recently continued to advance, most downstream applications such as object detection and semantic segmentation still employ ResNet variants as the backbone network due to their simple and modular structure. We present a simple and modular Split-Attention block that enables attention across feature-map groups. By stacking these Split-Attention blocks ResNet-style, we obtain a new ResNet variant which we call ResNeSt. Our network preserves the overall ResNet structure to be used in downstream tasks straightforwardly without introducing additional computational costs. ResNeSt models outperform other networks with similar model complexities. For example, ResNeSt-50 achieves 81.13 of 224x224, outperforming previous best ResNet variant by more than 1 accuracy. This improvement also helps downstream tasks including object detection, instance segmentation and semantic segmentation. For example, by simply replace the ResNet-50 backbone with ResNeSt-50, we improve the mAP of Faster-RCNN on MS-COCO from 39.3 from 42.1

READ FULL TEXT

page 1

page 2

page 3

page 4

research
05/30/2021

EPSANet: An Efficient Pyramid Split Attention Block on Convolutional Neural Network

Recently, it has been demonstrated that the performance of a deep convol...
research
01/27/2021

Bottleneck Transformers for Visual Recognition

We present BoTNet, a conceptually simple yet powerful backbone architect...
research
10/22/2020

Efficient Scale-Permuted Backbone with Learned Resource Distribution

Recently, SpineNet has demonstrated promising results on object detectio...
research
04/08/2022

Does Robustness on ImageNet Transfer to Downstream Tasks?

As clean ImageNet accuracy nears its ceiling, the research community is ...
research
03/28/2022

Optimal Correction Cost for Object Detection Evaluation

Mean Average Precision (mAP) is the primary evaluation measure for objec...
research
11/03/2022

Could Giant Pretrained Image Models Extract Universal Representations?

Frozen pretrained models have become a viable alternative to the pretrai...
research
09/08/2021

Panoptic SegFormer

We present Panoptic SegFormer, a general framework for end-to-end panopt...

Please sign up or login with your details

Forgot password? Click here to reset