AssafSinger94 / dino-tracker

Official Pytorch Implementation for “DINO-Tracker: Taming DINO for Self-Supervised Point Tracking in a Single Video”
MIT License
361 stars 39 forks source link

Training model for each different video #12

Closed tungyen closed 2 months ago

tungyen commented 2 months ago

For the training part, should I trained different model based on different video? I read the README file and found that each VIDEO_dir_path will save a model weight file. Thank you

tnarek commented 2 months ago

hi @tungyen. DINO-Tracker is an optimization-based method, meaning that it is trained only given a single video as input. So you are right, each video has its own weights.