mmdetection 的cascade_dcn_fpn50 转化为tensorrt后，采用本项目，发现转化成deepstream的模型比转化成tensorrt的模型精度较低，有些类阈值很低，有啥排查方法么

grimoire / mmdetection-to-tensorrt

convert mmdetection model to tensorrt, support fp16, int8, batch input, dynamic shape etc.

Apache License 2.0

590 stars 85 forks source link

Closed pangchao-git closed 2 years ago

pangchao-git commented 2 years ago

Describe the bug A clear and concise description of what the bug is.

To Reproduce Code snippet about how to reproduce the bug.

environment: Please provide your environment by:

python tools/collect_env.py

Please paste the result here.

Additional context Add any other context about the problem here.

grimoire commented 2 years ago

转换成deepstream是指什么？deepstream应该是直接使用TensorRT的模型的吧。 deepstream的预处理方式可能和mmdet中定义的有些不同，可以针对这一点做一些检查

pangchao-git commented 2 years ago

模型是rgb训练的,修改deepstream的配置文件设置输入是rgb,问题解决了