yhenon / pytorch-retinanet

Pytorch implementation of RetinaNet object detection.
Apache License 2.0
2.15k stars 665 forks source link

resize the image #59

Open wchzh opened 5 years ago

wchzh commented 5 years ago

I change the argument of "class Resizer(object):" . I change the "min_side=608, max_side=1024" to "min_side = 1600, max_side = 2400" . but I get error when I have training some epochs : [1] 1518 segmentation fault (core dumped). I do not know why this problem. Can you help me? Thank you.

gaopinghai commented 5 years ago

I have met the same error! I am running detection task with this beautiful model,and when one particular frame appears,the python code crashes with this fault:Segmentation fault (core dumped). In my opinion,it very possible that there are bugs in the C code NMS,hope the author can notice this. @yhenon

gaopinghai commented 5 years ago

I have met the same error! I am running detection task with this beautiful model,and when one particular frame appears,the python code crashes with this fault:Segmentation fault (core dumped). In my opinion,it very possible that there are bugs in the C code NMS,hope the author can notice this. @yhenon

Now the error disappeared, but I think I have done nothing......