When I use caffe, I will reshape the model use "net.blobs['data'].reshape" when the input is not equal to (22,22). So I can use any size of input data.
But when I use snpe with dlc file, I can only use input of size(22,22), I didn't find a method to reshape the model. and a variety of networks (e.g. fully convolutional ones) should be able to easily support variable input sizes regardless of the size used at model creation/conversion。
*/
If SNPE provided the ability to set the input size at runtime at network initialization (regardless of what size you used in model creation/conversion), I need to vary the input size at each inference .
I am also keen to find a solution to this problem.
I have same question, it seems like snpe can not support Non-Fixed-Size input, i think this is a bug, because tensorflow graph_transform is support this input type.
Especially for the object detection model I needed variable input shape [1,?,?,3]. how to generate dlc from such tensor shape?