yolo2-pytorch

yolo2-pytorch copied to clipboard

multi GPUs support?

2

when run with torch.nn.DataParallel(net).cuda(), there is "AttributeError: 'DataParallel' object has no attribute 'loss'". After I change `loss = net.loss` to `loss = net.module.loss`, there is a error "TypeError: unsupported operand...

psu1

could you please write a train demo?

2

According to the train.py file,I could know that :If I want to train my own data,I should write a class such as VOCDataset,but how can I do specifically?

gittigxuy

The issue about bbox loss and mAP when training my own data.

2

I only used one-class data to train the model and the change of loss is shown below.  Why is the bbox loss unable to converg? The class loss and...

applerjunjun

Hi, Why do I train your code on VOC2007 without darknet19.weights.npz only got 30% accuracy. Thanks a lot!

DW1HH

cann't download the yolo-voc.weights.h5](https://drive.google.com/open?id=0B4pXCfnYmG1WUUdtRHNnLWdaMEU)

3

I cann't download the yolo-voc.weights.h5](https://drive.google.com/open?id=0B4pXCfnYmG1WUUdtRHNnLWdaMEU) ,could anyone have it?Thank you!

deepblue0822

ln -s $VOCdevkit VOCdevkit2007

1

ln -s $VOCdevkit VOCdevkit2007 When I use this statement to process the data set, I found a locked file under the data folder, indicating that the resulting file is corrupted[...

zyyhl

I am curious about why the Anchor values in this [project ](https://github.com/longcw/yolo2-pytorch/blob/master/cfgs/config_voc.py#L15)are different from the [darknet](https://github.com/pjreddie/darknet/blob/master/cfg/yolov2-voc.cfg#L242)? And does this have impacts on the mAP?

ShoufaChen

i do not know how to deal it

hh99566

I have a question about the code in the function of def _process_batch in the darknet.py : the _boxes[:,:,0:2] = 0.5 _boxes[:,:,2:4] =1.0 .Thus the noobject will have the box...

zhengnianqing

when running demo.py , I did not get the output result in demo/out. Why? Thank you for you help!!!

chenhongming