search

z-x-yang / CFBI

The official implementation of CFBI(+): Collaborative Video Object Segmentation by (Multi-scale) Foreground-Background Integration.
BSD 3-Clause "New" or "Revised" License
322 stars 43 forks source link

how to make the 4-bit anotation image? #20

Closed mkmk001 closed 3 years ago

mkmk001 commented 3 years ago

I want to make my own dataset but I do not known how to get the 4-bit depth anotation.The PS software could only make the 8-bit ones.And when I use them,my GPU memory is not enouph.

z-x-yang commented 3 years ago

How many objects in your annotation?

The maximum value in the mask annotation should be the same as the number of containing objects. Since we use the maximum mask value to decide how many objects are in the corresponding video sequence.

If the maximum value is 5, for example, and there is only 1 object in the video. The inference code will predict this object with 4 empty objects (because they don't have any mask region), and you will need much more unnecessary GPU memory.

Set the mask value to be 1 for the first object, to be 2 for the second, and so on.

z-x-yang commented 3 years ago

It seems that the problem has been solved. I'll close the issue.