search

Vegeta2020 / SE-SSD

SE-SSD: Self-Ensembling Single-Stage Object Detector From Point Cloud, CVPR 2021.
Apache License 2.0
811 stars 128 forks source link

Problem during training #79

Closed badalpardhi12 closed 2 years ago

badalpardhi12 commented 2 years ago

I having an error when I run train.py with pretrained CIA-SSD model. Any help would be greatly appreciated! Also, any tips on using this model on a custom dataset with 16 channel lidar data.

image

Vegeta2020 commented 2 years ago

@badalpardhi12 From above traceback, there is no points selected for concatenation, and you may check if the class_names are set correctly. Also, it' easy to change the input lidar data from current 4 channels to 16 channels, you may do it in the config file, but you have to check the data preprocessing files in pipeline carefully to avoid any possible inconsistency.