Self-distillation for video

OpenGVLab / VideoMamba

[ECCV2024] VideoMamba: State Space Model for Efficient Video Understanding

https://arxiv.org/abs/2403.06977

Apache License 2.0

844 stars 60 forks source link

Self-distillation for video #72

Open isacciobota opened 4 months ago

isacciobota commented 4 months ago

Is self-distillation only implemented for image processing model? videomamba_distill.py is only in the image_sm/models folder. I would like to use videoMamba as a backbone for video 3D pose estimation but not sure if self-distillation works for video.

Andy1621 commented 4 months ago

Hi! I only use self-distillation for image classification to avoid the overfitting issues, since it runs normal for video tasks.

YCHYZW commented 4 weeks ago

Can you provide pre-training weights for clip in single-modal video training?