When using the model trained by myself, the following problems are encountered. Is it due to the process of onnx transfer?What changes should I make to the code when the number of categories changes?

NVIDIA-AI-IOT / CUDA-PointPillars

A project demonstrating how to use CUDA-PointPillars to deal with cloud points data from lidar.

Apache License 2.0

502 stars 148 forks source link

When using the model trained by myself, the following problems are encountered. Is it due to the process of onnx transfer?What changes should I make to the code when the number of categories changes? #88

Open NukaCC opened 1 year ago

NukaCC commented 1 year ago

Hello, I trained my model on openpcdet. The model I trained has a total of 8 classes and was successfully converted to onnx. However, the following problems occurred when running demo:

MatMul_172: last dimension of inpute = 16 and second to last dimension of input1 = 10 but must match.failed to parse onnx model file, please check the onnx version and trt support op!

ACFIRSTONE commented 7 months ago

lidar-postprocess.hpp simplify_onnx.py(simplify_postprocess)