search

zlai0 / MAST

MAST: A Memory-Augmented Self-supervised Tracker (CVPR 2020)
https://zlai0.github.io/MAST/
273 stars 32 forks source link

How to process memories frame by frame #5

Open ruoqi77 opened 4 years ago

ruoqi77 commented 4 years ago

Sorry to bother. Another question. So far, my experience in coding is not so much. I want to reduce the memories by changing codes to process frame by frame but don't know how. Could you please give me some hints? Thank you so much!

zlai0 commented 4 years ago

You can add an extra option --ref 2 when you test.

ruoqi77 commented 4 years ago

You can add an extra option --ref 2 when you test.

But this will make the performance drop, I mean how to process memories frame by frame while preserving the accuracy, like you said in readme.

ruoqi77 commented 4 years ago

So sorry to bother, I have another question. In model/colorizer.py lines 45-46, I fail to figure out why do you add one-hot operation in inference stage？Looking forward to your reply！

zlai0 commented 3 years ago

This is to enable multi-object tracking in a single inference's time (DAVIS, Youtube-VOS). Basically each one-hot label represents an instance label.