Error when use inception_v3 model

1

i want to use the inception_v3 model, and my input is [128,3,44,44], i want to classifier 7.

122
1221971×177 19 KB
QQ截图20200108221039
QQ截图202001082210393337×877 323 KB

2

Your input tensor is too small for this architecture and at some point the activation output will be an empty tensor.
The Inception model expects your output to have a spatial size of 299x299.
You could use torchvision.transform.Resize to increase the spatial size of your images.

3

thanks so much :smiley: