Hi all,
I was planning on training a model that requires 640x640 of input image size, but the base models eIQ software only supports 320x320 input size. So I decided to modify the input layer of the base model to 640x640 such that I can avoid training it with 320x320(the base model is "C:\nxp\eIQ_Toolkit_v1.0.5\plugins\detection\boxes\fpn\VOC-fpn_ssd_mobilenet_v2-320.h5"). Is this a recommended approach in general and also how can I achieve this goal? Thanks very much!