search

zlai0 / MAST

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

In table 5,how many reference frames in only short? #12

Open houhouhouhou11 opened 3 years ago

houhouhouhou11 commented 3 years ago

Hello: Thanks for your perfect work. In table 5,memory is divided into only short, only long, and mixed .I want to know how many reference frames in only short to get the result J-mean 0.573 F-mean 0.618 ? thank you very much !(^_^)

khai9xht commented 3 years ago

I have same question, because when I evaluate with 1 reference, i have bad result. I demo 2 sequence frame below image image

houhouhouhou11 commented 3 years ago

@khai9xht hello, I have meet the same question. In testing process, do you know why we should prep the quantized_r to one-hot format with 7 channels in the 46 line of the colorizer.py ? Thank you!

zlai0 commented 3 years ago

@houhouhouhou11 "Only short" should be 1 frame. During testing process each label is propagated separately (just like each of three color channels).

zlai0 commented 3 years ago

I have same question, because when I evaluate with 1 reference, i have bad result. I demo 2 sequence frame below

Could you first try evaluate J&F to see if you get similar number as reported in the paper? There are indeed some hard sequences.