Some problems about DIou-darknet-tiny.

[y200504040u]

I trained my tiny version by the implement. I added the ciou parameters in each yolo layer of yolo-tiny.cfg in the original Darknet based on the coco-ciou.cfg like this.

My training log like this:

I trained only person class.The mean iou is above 0.5. I run inference on the test image by my weight, the result is no box.

I wanted to make sure that I wasn't doing something wrong.

[Zzh-tju]

You want to open the picture on the server, right? The image is on yourpath/DIoU-Darknet/prediction.jpg

[y200504040u]

You want to open the picture on the server, right? The image is on yourpath/DIoU-Darknet/prediction.jpg

Yes.I download the prediction.jpg from the server to my PC.The result is none. The log is like this if I get result.

I log the num of boxes like this.Then make

But got none. Maybe there is something wrong in network running.

[Zzh-tju]

can you show me your cfgfile？ @y200504040u

[y200504040u]

can you show me your cfgfile？ @y200504040u

[net]

Testing

batch=1

subdivisions=1

Training

batch=256 subdivisions=8 width=608 height=608 channels=3 momentum=0.9 decay=0.0005 angle=0 saturation = 1.5 exposure = 1.5 hue=.1

single gpu

learning_rate=0.001

burn_in=1000

max_batches = 550400

2 gpu

learning_rate=0.0005

burn_in=2000

max_batches = 500200

4 gpu

learning_rate=0.00025 burn_in=4000 max_batches = 500200

learning_rate=0.001

burn_in=1000

max_batches = 500200

policy=steps steps=400000,450000 scales=.1,.1

[convolutional] batch_normalize=1 filters=16 size=3 stride=1 pad=1 activation=leaky

[maxpool] size=2 stride=2

[convolutional] batch_normalize=1 filters=32 size=3 stride=1 pad=1 activation=leaky

[maxpool] size=2 stride=2

[convolutional] batch_normalize=1 filters=64 size=3 stride=1 pad=1 activation=leaky

[maxpool] size=2 stride=2

[convolutional] batch_normalize=1 filters=128 size=3 stride=1 pad=1 activation=leaky

[maxpool] size=2 stride=2

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[maxpool] size=2 stride=2

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[maxpool] size=2 stride=1

[convolutional] batch_normalize=1 filters=1024 size=3 stride=1 pad=1 activation=leaky

###########

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[convolutional] size=1 stride=1 pad=1 filters=18 activation=linear

[yolo] mask = 3,4,5 anchors = 10,14, 23,27, 37,58, 81,82, 135,169, 344,319 classes=1 num=6 jitter=.3 ignore_thresh = .7 truth_thresh = 1 random=1 iou_normalizer=0.5 cls_normalizer=1.0 iou_loss=ciou nms_kind=greedynms beta1=0.6

[route] layers = -4

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[upsample] stride=2

[route] layers = -1, 8

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[convolutional] size=1 stride=1 pad=1 filters=18 activation=linear

[yolo] mask = 0,1,2 anchors = 10,14, 23,27, 37,58, 81,82, 135,169, 344,319 classes=1 num=6 jitter=.3 ignore_thresh = .7 truth_thresh = 1 random=1 iou_normalizer=0.5 cls_normalizer=1.0 iou_loss=ciou nms_kind=greedynms beta1=0.6

[Zzh-tju]

maybe you should test on another weight file

[y200504040u]

Could you give me your email? Thanks.

[Zzh-tju]

zh_zheng@tju.edu.cn