search

DerrickXuNu / OpenCOOD

[ICRA 2022] An opensource framework for cooperative detection. Official implementation for OPV2V.
https://mobility-lab.seas.ucla.edu/opv2v/
Other
663 stars 99 forks source link

Models provided in the link seems broken #123

Closed drowning-in-codes closed 9 months ago

drowning-in-codes commented 10 months ago

Thanks for your codebase! I tested the model provided in PointPillar_fcooper, but the inference speed seems strangely slow and the result is weird. I run the command below to show the result, it seems like pred_boxs in red are different from gt_box in green. python opencood/tools/inference.py --model_dir "./opencood/logs/pointpillar_fcooper" --fusion_method "intermediate" --save_vis image and the ap is also weird. image I think there's something wrong with this model or is there anything wrong with me?

2429993608 commented 10 months ago

"How is your visualization configured, and why am I unable to visualize it?"

drowning-in-codes commented 10 months ago

"How is your visualization configured, and why am I unable to visualize it?"

I don't know your setup, but I used remote server for infer and found that it needs an additional settings in the remote server to visualize(Iit's related to VNC and some stuff). I notice that open3d is used to visualize and it needs to create a GUI window which is not supported in non-a GUI remote server without additional settings.So I train on remote and inference with visualization local.

2429993608 commented 10 months ago

"How is your visualization configured, and why am I unable to visualize it?"

I don't know your setup, but I used remote server for infer and found that it needs an additional settings in the remote server to visualize(Iit's related to VNC and some stuff). I notice that open3d is used to visualize and it needs to create a GUI window which is not supported in non-a GUI remote server without additional settings.So I train on remote and inference with visualization local.

ok~ Thank you very much for your reply.

2429993608 commented 10 months ago

"How is your visualization configured, and why am I unable to visualize it?"

I don't know your setup, but I used remote server for infer and found that it needs an additional settings in the remote server to visualize(Iit's related to VNC and some stuff). I notice that open3d is used to visualize and it needs to create a GUI window which is not supported in non-a GUI remote server without additional settings.So I train on remote and inference with visualization local.

Hello, could you please tell me about your environment configuration? My configuration is as follows: pytorch 1.12.0 Spconv 2.x (spconv-cu102-2.3.6) open3d 0.17.0

drowning-in-codes commented 9 months ago

"How is your visualization configured, and why am I unable to visualize it?"

I don't know your setup, but I used remote server for infer and found that it needs an additional settings in the remote server to visualize(Iit's related to VNC and some stuff). I notice that open3d is used to visualize and it needs to create a GUI window which is not supported in non-a GUI remote server without additional settings.So I train on remote and inference with visualization local.

Hello, could you please tell me about your environment configuration? My configuration is as follows: pytorch 1.12.0 Spconv 2.x (spconv-cu102-2.3.6) open3d 0.17.0

I just follow the installation guide， clone the repo, and type the command below, so the dependency is totally the same as in equirements.txt, and python version is 3.7. conda env create -f environment.yml conda activate opencood python setup.py develop pip install -r requirements.txt

drowning-in-codes commented 9 months ago

well, it looks like I have got it wrong.