Home

Awesome

SwinT_detectron2

Swin Transformer for Object Detection by detectron2

This repo contains the supported code and configuration files to reproduce object detection results of Swin Transformer. It is based on detectron2.

You can find SwinV2 in this repo

Results and Models

RetinaNet

BackbonePretrainLr Schdbox mAPmask mAP#paramsFLOPsconfiglogmodel
Swin-TImageNet-1K3x44.6---config-model

Faster R-CNN

BackbonePretrainLr Schdbox mAPmask mAP#paramsFLOPsconfiglogmodel
Swin-T FPNImageNet-1K3x45.1---config-model

Mask R-CNN

BackbonePretrainLr Schdbox mAPmask mAP#paramsFLOPsconfiglogmodel
Swin-T FPNImageNet-1K3x45.541.8--config-model

The mask mAP (41.8 vs 41.6) is same as the mmdetection, but box mAP is worse (45.5 vs 46.0)

Usage

Please refer to get_started.md for installation and dataset preparation.

note: you need convert the original pretrained weights to d2 format by convert_to_d2.py

References