Hello there! I have a few questions about spp_net. I look forward to your answers

[doublesmallcw]

1. Using the pre-trained network to add the spp layer before the final fully connected layer, can I still use the parameters of the pre-trained network for migration learning?
   2 . I saw that your spp_ResNet_50 network has also performed ReSize(224,224). Isn’t it possible to use the SPP layer without resize? SPP can handle pictures of various specifications, right?
   3 When adding SPP to the pytorch pre-training network, can I not re-construct the network structure? Is it possible to just add an SPP layer to the pre-trained network?