yjh0410 87d5a72477 modify init		1 rok temu
..
README.md	7baf8b2382 release YOLOv2	1 rok temu
build.py	264112178f build a new YOLO-Tutorial project for my book	1 rok temu
loss.py	264112178f build a new YOLO-Tutorial project for my book	1 rok temu
matcher.py	264112178f build a new YOLO-Tutorial project for my book	1 rok temu
resnet.py	264112178f build a new YOLO-Tutorial project for my book	1 rok temu
yolov2.py	264112178f build a new YOLO-Tutorial project for my book	1 rok temu
yolov2_backbone.py	264112178f build a new YOLO-Tutorial project for my book	1 rok temu
yolov2_basic.py	87d5a72477 modify init	1 rok temu
yolov2_head.py	264112178f build a new YOLO-Tutorial project for my book	1 rok temu
yolov2_neck.py	87d5a72477 modify init	1 rok temu
yolov2_pred.py	606c352336 modify pred init	1 rok temu

Redesigned YOLOv2:

Model	Backbone	Batch	Scale	AP^val 0.5	Weight	Logs
YOLOv2	ResNet-18	1xb16	640	75.7	ckpt	log

For training, we train redesigned YOLOv2 with 150 epochs on COCO.
For data augmentation, we use the SSD's augmentation, including the RandomCrop, RandomDistort, RandomExpand, RandomHFlip and so on.
For optimizer, we use AdamW with weight decay of 0.05 and per image base lr of 0.001 / 64.
For learning rate scheduler, we use cosine decay scheduler.
For batch size, we set it to 16, and we also use the gradient accumulation to approximate batch size of 256.

Train YOLOv2

Taking training YOLOv2-R18 on COCO as the example,

python train.py --cuda -d coco --root path/to/coco -m yolov2_r18 -bs 16 --fp16

Taking training YOLOv2-R18 on COCO as the example,

python -m torch.distributed.run --nproc_per_node=8 train.py --cuda --distributed -d coco --root path/to/coco -m yolov2_r18 -bs 16 --fp16

Taking testing YOLOv2-R18 on COCO-val as the example,

python test.py --cuda -d coco --root path/to/coco -m yolov2_r18 --weight path/to/yolov2.pth --show

Taking evaluating YOLOv2-R18 on COCO-val as the example,

python eval.py --cuda -d coco --root path/to/coco -m yolov2_r18 --weight path/to/yolov2.pth

python demo.py --mode image --path_to_img path/to/image_dirs/ --cuda -m yolov2_r18 --weight path/to/weight --show

python demo.py --mode video --path_to_vid path/to/video --cuda -m yolov2_r18 --weight path/to/weight --show --gif

python demo.py --mode camera --cuda -m yolov2_r18 --weight path/to/weight --show --gif

Model	Backbone	Batch	Scale	AP^val 0.5:0.95	AP^val 0.5	FLOPs ^(G)	Params ^(M)	Weight	Logs
YOLOv2	ResNet-18	1xb16	640	28.4	47.4	38.0	21.5	ckpt	log