MEYER Adrien

Results 5 comments of MEYER Adrien

Could mmtracking support anchor-free detectors for VID

> SELSA only supports two-stage detectors. This is because the SELSA module is design to aggregate proto detection; it weights the proto detection of reference frames based on target frames...

cannot reshape tensor of 0 elements into shape [0, 16, -1] because the unspecified dimension size -1 can be any value and is ambiguous

are you up to date with #463 ?

[Feature] Incorporating TF-Blender for VOD

> 1. Incorporated [TF-Blender ](https://openaccess.thecvf.com/content/ICCV2021/papers/Cui_TF-Blender_Temporal_Feature_Blender_for_Video_Object_Detection_ICCV_2021_paper.pdf) into Video Object Detection methods. > > 2. The source code is borrowed from the original authors ([HERE](https://github.com/goodproj13/TF-Blender)) > > 3. The module TF-Blender is...

Multi-GPU training hangs

Same issue, only with multi gpu

Grid search YOLOX base_lr

Yes, sometime the gridsearch require inter dependant values which is currently not supported. I would like this feature as well.