Hello, I have applied post training static quantized the ssdlite model from torchvision. The model has been successfully quantized, however when I try to infer the model, the following error comes up.
“in Normalize raise TypeError( TypeError: Expected input images to be of floating type (in range [0, 1]), but found type torch.quint8 instead”
I can understand that the transform in the torchvision library is the reason of this issue. However, I couldn’t fix this issue just by commenting out “transform part” from my local library file.
The issue on this part can be solved by not applying QuantStub but model part gives another error.
I’d like to ask the way to successfully quantize ssdlite model from torchvision.