search

ayooshkathuria / YOLO_v3_tutorial_from_scratch

Accompanying code for Paperspace tutorial series "How to Implement YOLO v3 Object Detector from Scratch"
https://blog.paperspace.com/how-to-implement-a-yolo-object-detector-in-pytorch/
2.32k stars 724 forks source link

RuntimeError in util.py : torch.cuda.Floattensor #26

Closed siddharthkale97 closed 5 years ago

siddharthkale97 commented 5 years ago

In util.py file I am getting an error at the line

prediction[:,:,:2] += x_y_offset

saying that

Traceback (most recent call last): File "darknet.py", line 239, in pred = model(inp, torch.cuda.is_available()) File "C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 477, in call result = self.forward(*input, **kwargs) File "darknet.py", line 225, in forward x = predict_transform(x, inp_dim, anchors, num_classes, CUDA) File "C:\Users\Siddharth Kale\Desktop\Coding Essentials\codes\Convoulution_Neural_Nets\yolo-pytorch\util.py", line 43, in predict_transform prediction[:,:,:2] += x_y_offset RuntimeError: Expected object of type torch.FloatTensor but found type torch.cuda.FloatTensor for argument #4 'other'

Couldn't find any solutions, could someone suggest something to get rid of this error``

AvivSham commented 5 years ago

I also encountered the same problem, I have tried to change a few things without success. Help, please.

I think I found the source of the problem. When you use: print(x_y_offset.device) print(prediction.device)

you get cuda:0 for x_y_offset and cpu for prediction! I just don't know how to solve this. I'm using colab notebook.

AvivSham commented 5 years ago

OK, I got it! (works for me) After prediction[:,:,4] = torch.sigmoid(prediction[:,:,4]) add: prediction = prediction.cuda()

without the added line prediction is calculated in the CPU thus the problem!

Please keep me posted if it solves your problem.

siddharthkale97 commented 5 years ago

OK, I got it! (works for me) After prediction[:,:,4] = torch.sigmoid(prediction[:,:,4]) add: prediction = prediction.cuda()

without the added line prediction is calculated in the CPU thus the problem!

Please keep me posted if it solves your problem.

I tried with adding prediction = prediction.cuda() but I kept on getting the error. But I aslo encountered a strange thing. While following the tutorial I got this problem, but strangely the problem was solved when I cloned the repo and then run the program on a test timage. Anyway thanks for your help