Dongming Wu

Results 21 comments of Dongming Wu

I am not sure if the script "visualize_results.py" is available. I visualize my results from "test.py" in terms of Openlanev2: > Then you can train or evaluate a model using...

https://github.com/OpenDriveLab/OpenLane-V2/blob/master/docs/getting_started.md?plain=1#L71

Hi, I recommend you to use our new repo: https://github.com/zyn213/TempRMOT

Please refer to https://github.com/wudongming97/TopoMLP/blob/master/docs/setup.md

https://github.com/OpenDriveLab/OpenLane-V2/blob/master/docs/getting_started.md?plain=1#L71

Please refer to https://github.com/wudongming97/Prompt4Driving/blob/master/projects/tracking_plugin/models/trackers/tracker_prompt.py#L353

In the beginning, we use bezier curves because they are visibly smooth. However, from our latest experiments, regressing 11 points performs similarly to the bezier method.

We used a very early version of flash-attention: https://github.com/Dao-AILab/flash-attention/tree/v0.2.2 You can try to install it locally.

I used 8*2080ti for model training.

You need to build MultiScaleDeformableAttention. Please see MOTR for this building: https://github.com/megvii-research/MOTR