Hi,
I am trying to run the classification model using object detection usecase. I replaced the NET_OD.BIN and PRM_OD.BIN with classification network and parameter. Based on the classification output from the model, I will change the position of box being drawn on the image.
I am getting the pointer to the output of softmax buffer on ObjectDrawPlugin as follows:
metaDataAddr = pCompositeBuffer->bufAddr[0][1]; //ChId 0 contains the image.
bufferOffset = ((44+8)*4) + 4; // Due to the padding of 4 on the borders
classPtr = ((UInt8 *)((char *)metaDataAddr)) + bufferOffset;
But the classid I'm getting is not proper. How to get the classification output at the next layer?
When I tried to run the model using TIDL FILE IO Usecase it dumped 2304 bytes instead of 44bytes(44 class model).
Regards,
Navinprashath.R.R