Segmentation fault (core dumped) about ssds.pytorch HOT 4 CLOSED

hello，can you tell me how to solve this problem，i have a same problem... thanks

from ssds.pytorch.

@QZ-cmd @ParusMajor60
Please try the newest pytorch version (1.5) with dev branch and check whether the problem still existed. Also it is good to attach the detail log for better debugging.

from ssds.pytorch.

@QZ-cmd @ParusMajor60
Please try the newest pytorch version (1.5) with dev branch and check whether the problem still existed. Also it is good to attach the detail log for better debugging.

Thank you for your reply，i change : images = torch.Tensor(images).to(self.device) -> images = images.to(self.device) in yolov2_test_model.py solve this problem,but I don't know what the negative impact will be，Look forward to your reply

from ssds.pytorch.

Based on the given code, the images = torch.Tensor(images).to(self.device) expect for the numpy.ndarray input and transfer it to the torch.Tensor, while images = images.to(self.device) seems already get the torch.Tensor for images. Since I'm not so sure where does the yolov2_test_model.py script come from. So not so sure about the potential issue for that script.

Btw, welcome to try the new 1.5 version for ssds.pytorch. I'm working on the document in the recent days and it is available at https://foreveryounggithub.github.io/ssds.doc/

from ssds.pytorch.