search

NotACracker / COTR

[CVPR24] COTR: Compact Occupancy TRansformer for Vision-based 3D Occupancy Prediction
Apache License 2.0
47 stars 4 forks source link

How COTR is integrated with the baselines #4

Closed LeeKeyu closed 3 months ago

LeeKeyu commented 3 months ago

Thanks for your great and impressive work! I am wondering if you can elaborate more on how the COTR is integrated with several baseline methods BEVDet4D [8], TPVFormer [10], SurroundOcc [37] and OccFormer [41], as this seems not clearly described in the paper or code. During reading the paper, I thought COTR itself was an alternative to the baseline methods, e.g., TPVFormer, SurroundOcc, which shared the same input and output as COTR. So I was really confused by the statement that the COTR was integrated with the aforementioned methods to obtain the results shown in the paper. I suppose that it would be very helpful to understand the information flow and changes of these methods by integrating COTR. Thanks a lot in advance! Your advice will of great help to me and other potential readers.

NotACracker commented 3 months ago

Thanks for your great and impressive work! I am wondering if you can elaborate more on how the COTR is integrated with several baseline methods BEVDet4D [8], TPVFormer [10], SurroundOcc [37] and OccFormer [41], as this seems not clearly described in the paper or code. During reading the paper, I thought COTR itself was an alternative to the baseline methods, e.g., TPVFormer, SurroundOcc, which shared the same input and output as COTR. So I was really confused by the statement that the COTR was integrated with the aforementioned methods to obtain the results shown in the paper. I suppose that it would be very helpful to understand the information flow and changes of these methods by integrating COTR. Thanks a lot in advance! Your advice will of great help to me and other potential readers.

We have provided configuration files for integrating COTR into SurroundOCC or Occformer, which can be found in configs/cotr/.

LeeKeyu commented 3 months ago

ok thanks. I will look into the code.