search

yichen928 / SparseFusion

[ICCV 2023] SparseFusion: Fusing Multi-Modal Sparse Representations for Multi-Sensor 3D Object Detection
Apache License 2.0
198 stars 19 forks source link

The problem with FPS during inferencing #24

Closed liyih closed 10 months ago

liyih commented 10 months ago

Hi, yichen, Thanks for your awesome work. I use A100 to inference your Resnet-50 based model, and only get FPS=2.1. It's strange, as in your table 4, you got FPS=5.6 with A6000. I use tools/analysis_tools/benchmark.py for inference. I want to know how do you calculate the FPS in your model. Best

liyih commented 10 months ago

I guess you don't consider voxelization time just as bevfusion(MIT) do. I remove the voxelization time and got fps=6.4 in A100, that looks good.

yichen928 commented 10 months ago

I guess you don't consider voxelization time just as bevfusion(MIT) do. I remove the voxelization time and got fps=6.4 in A100, that looks good.

That is correct! Thanks for your interest.