PaddleDetection/configs/detr

• DETR
• • Introduction
  • Model Zoo
  • Citations

DETR

Introduction

DETR is an object detection model based on transformer. We reproduced the model of the paper.

Model Zoo

Backbone	Model	Images/GPU	Inf time (fps)	Box AP	Config	Download
R-50	DETR	4	---	42.3	config	model

Notes:

• DETR is trained on COCO train2017 dataset and evaluated on val2017 results of mAP(IoU=0.5:0.95).
• DETR uses 8GPU to train 500 epochs.

GPU multi-card training

export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
python -m paddle.distributed.launch --gpus 0,1,2,3,4,5,6,7 tools/train.py -c configs/detr/detr_r50_1x_coco.yml --fleet -o find_unused_parameters=True

Citations

@inproceedings{detr,
  author    = {Nicolas Carion and
               Francisco Massa and
               Gabriel Synnaeve and
               Nicolas Usunier and
               Alexander Kirillov and
               Sergey Zagoruyko},
  title     = {End-to-End Object Detection with Transformers},
  booktitle = {ECCV},
  year      = {2020}
}