Anyone can provide a detailed tutorial of how to convert FasterRCNN caffe models to FasterRCNN SNPE models ? Specifically, how can I set the PRE_NMS_TOP_N and POST_NMS_TOP_N for the Proposal Layer ?
Also, R-FCN is a very popular improvement on FasterRCNN. It would be great if future SNPE revisions support R-FCN-based object detection algorithms.
I have converted Faster-RCNN caffe models to dlc models. You should provide the params of proposal layer in Python type.