yolov7

Implementation of "YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors"

This implimentation is for darknet.

Usage

yolov7-tiny.weights yolov7.weights yolov7x.weights

./darknet detector demo cfg/coco.data cfg/yolov7-tiny_darknet.cfg yolov7-tiny.weights test.mp4
./darknet detector demo cfg/coco.data cfg/yolov7_darknet.cfg yolov7.weights test.mp4
./darknet detector demo cfg/coco.data cfg/yolov7x_darknet.cfg yolov7x.weights test.mp4

ELAN

Model	Size	APval	AP50val	AP75val	#Param.	FLOPs	Setting
ELAN-tiny (nyc)	640	36.2%	53.6%	39.5%	3.9M	16.1G	{1,1}
ELAN-tiny	640	41.1%	58.2%	44.7%	4.9M	19.4G	{2,1}
ELAN-tiny	640	42.1%	58.2%	45.7%	5.2M	20.4G	{1,2}
ELAN-tiny	640	44.5%	60.9%	48.2%	6.2M	23.6G	{2,2}
Dark-ELAN-tiny	640	44.4%	60.8%	48.0%	5.6M	21.8G	{1,3}
Dark-ELAN-tiny	640	45.0%	61.3%	48.6%	6.2M	23.6G	{3,3}
Res-ELAN-tiny	640	44.3%	60.7%	48.0%	5.4M	21.1G	.
CSP-ELAN-tiny	640	%	%	%	5.6M	21.6G	{1,3,1}
CSP-ELAN-tiny	640	%	%	%	5.9M	22.7G	{1,3,2}
CSP-ELAN-tiny	640	45.4%	61.9%	49.1%	5.9M	22.5G	{3,3,1}
CSP-ELAN-tiny (relu)	640	44.1%	60.6%	47.7%	5.9M	22.5G	{3,3,1}
CSP-ELAN-tiny	640	%	%	%	6.5M	24.5G	{3,3,2}
DarkN-ELAN-tiny	640	42.1%	58.6%	45.8%	4.4M	17.9G	{1,3}
DarkN-ELAN-tiny	640	43.5%	59.8%	47.3%	5.0M	19.7G	{3,3}
ResN-ELAN-tiny	640	41.9%	58.4%	45.8%	4.2M	17.1G	.
CSPN-ELAN-tiny	640	43.7%	60.2%	47.5%	4.6M	18.6G	{3,3,1}
CSPN-ELAN-tiny	640	44.7%	61.1%	48.5%	5.2M	20.5G	{3,3,2}