search

Little-Podi / GRM

[CVPR'23] The official PyTorch implementation of our CVPR 2023 paper: "Generalized Relation Modeling for Transformer Tracking".
MIT License
69 stars 8 forks source link

test speed #10

Closed June-hui closed 1 year ago

June-hui commented 1 year ago

hello ! i want to know how you test speed on your model (45fps on got10k)? can u provide the "profile_model.py" like other tracker? thank u!

Little-Podi commented 1 year ago

Hi, I directly took the average FPS printed during inferecne as the running speed of my tracker. The profile_model.py needs the inputs manually prepared, and I do have a messy version. It would be an useful reference, whereas normally it will provide a faster result. Besides, the amount of FLOPs also cannot truly reveal the computation latency of algorithms. The most simple and straightforward way is still running them on the same machine. If you want it, I think you can smoothly modify a script from STARK or OSTrack.