Gigalomanicx

Results 3 issues of Gigalomanicx

没有使用tools/analysis_tools/vis.py,而是直接从result_nusc.json中选取pred_bbox绘制。但是出现了检测框错位现象,检查了很久。 ![413893c8a01444035fa5b8fab87b14a7](https://github.com/HuangJunJie2017/BEVDet/assets/109480200/7e80f830-cb39-461e-b5a9-26f8c548defd) 不太确定是因为数据增强还是测试时的'MultiScaleFlipAug3D'。有人遇到过这种情况吗?

When I try`python tools/test.py $configs $checkpoint --format-only --eval-options jsonfile_prefix=jsonsave/` The following error appears to be looking for annotations? ![image](https://github.com/HuangJunJie2017/BEVDet/assets/109480200/ab16d942-ef27-4a51-9c29-6f189f8e449f) When I remove the config file: ![image](https://github.com/HuangJunJie2017/BEVDet/assets/109480200/7e0caf0a-9714-4101-8b0f-43541cd3638f) Then: ![image](https://github.com/HuangJunJie2017/BEVDet/assets/109480200/40f5b6f4-ce45-4d9f-9d5e-d16fe61fb8cb) I'm not...

I am currently working on extending the LiDAR modality within the StreamPETR framework. However, I am unsure how to incorporate historical frame **sweep points** into the **data{}** during the forward_train...