models/official/projects/detr

End-to-End Object Detection with Transformers (DETR)

.

TensorFlow 2 implementation of End-to-End Object Detection with Transformers

⚠️ Disclaimer: All datasets hyperlinked from this page are not owned or distributed by Google. The dataset is made available by third parties. Please review the terms and conditions made available by the third parties before using the data.

Scripts:

You can find the scripts to reproduce the following experiments in detr/experiments.

DETR COCO (ImageNet pretrained)

Model	Resolution	Batch size	Epochs	Decay@	Params (M)	Box AP	Dashboard	Checkpoint	Experiment
DETR-ResNet-50	1333x1333	64	300	200	41	40.6	tensorboard	ckpt	detr_r50_300epochs.sh
DETR-ResNet-50	1333x1333	64	500	400	41	42.0	tensorboard	ckpt	detr_r50_500epochs.sh
DETR-ResNet-50	1333x1333	64	300	200	41	40.6	paper	NA	NA
DETR-ResNet-50	1333x1333	64	500	400	41	42.0	paper	NA	NA
DETR-DC5-ResNet-50	1333x1333	64	500	400	41	43.3	paper	NA	NA

Need contribution:

• Add DC5 support and update experiment table.

Citing TensorFlow Model Garden

If you find this codebase helpful in your research, please cite this repository.

@misc{tensorflowmodelgarden2020,
  author = {Hongkun Yu and Chen Chen and Xianzhi Du and Yeqing Li and
            Abdullah Rashwan and Le Hou and Pengchong Jin and Fan Yang and
            Frederick Liu and Jaeyoun Kim and Jing Li},
  title = {{TensorFlow Model Garden}},
  howpublished = {\url{https://github.com/tensorflow/models}},
  year = {2020}
}