History

Xiaolin Wang 71e5c793e5 [Doc] Fix typo (#270 ) * fix typo * fix ci		2022-11-10 11:12:31 +08:00
..
cspnext_imagenet_pretrain	[Feature] Support RTMDet (#85 )	2022-09-29 19:09:56 +08:00
README.md	[Doc] Fix typo (#270 )	2022-11-10 11:12:31 +08:00
metafile.yml	[Fix] Update metafile for RTMDet (#188 )	2022-11-03 19:03:06 +08:00
rtmdet_l_syncbn_8xb32-300e_coco.py	[Feature] Support P6 YOLOv5 (#168 )	2022-11-03 19:03:06 +08:00
rtmdet_m_syncbn_8xb32-300e_coco.py	[Feature] Support RTMDet (#85 )	2022-09-29 19:09:56 +08:00
rtmdet_s_syncbn_8xb32-300e_coco.py	Add changelog of v0.1.2 (#226 )	2022-11-03 19:03:06 +08:00
rtmdet_tiny_syncbn_8xb32-300e_coco.py	[Feature] Support RTMDet (#85 )	2022-09-29 19:09:56 +08:00
rtmdet_x_syncbn_8xb32-300e_coco.py	[Feature] Support RTMDet (#85 )	2022-09-29 19:09:56 +08:00

README.md

RTMDet

Abstract

Our tech-report will be released soon.

Results and Models

Backbone	size	SyncBN	box AP	Params(M)	FLOPS(G)	TRT-FP16-Latency(ms)	Config	Download
RTMDet-tiny	640	Yes	40.9	4.8	8.1	0.98	config	model \| log
RTMDet-s	640	Yes	44.5	8.89	14.8	1.22	config	model \| log
RTMDet-m	640	Yes	49.1	24.71	39.27	1.62	config	model \| log
RTMDet-l	640	Yes	51.3	52.3	80.23	2.44	config	model \| log
RTMDet-x	640	Yes	52.6	94.86	141.67	3.10	config	model \| log

Note:

The inference speed is measured on an NVIDIA 3090 GPU with TensorRT 8.4.3, cuDNN 8.2.0, FP16, batch size=1, and without NMS.
We still directly use the weights trained by mmdet currently. A re-trained model will be released later.