Regarding the issue of inconsistent inference results between the openpcdet model and the generated plan model

[hehualin-tut]

Problem description: I trained a custom dataset using openpcdet, and because the dataset contains the same object at different heights, I generated anchors for the same object at different heights, as shown in the following figure

The results of using the PT model for inference are also quite in line with my expectations

But when I convert to the plan model, the detection results will bounce back and forth, unlike the stability detected by the PT model

This is the image I generated for onnx and the related modifications in post-processing. I would like to know if I need to modify any other files?