Add multiple-object segmentation feature and fine-tuning on multiple images

[mlzoo]

Add multiple-object segmentation feature and fine-tuning on multiple images.

[OSSome01]

Is this for multiple-object segmentation of the same class? or is it also for multiple classes?

[mlzoo]

Is this for multiple-object segmentation of the same class? or is it also for multiple classes?

This is for multiple objects of the same class appearing in the one image.

[Skwarson96]

Hi! @mlzoo is your code working? I found some syntax errors and many more errors, mainly: "local variable ... referenced before assignment"

[ZrrSkywalker]

@mlzoo Hi, thanks very much for your contribution! It means a lot for PerSAM to tackle multi-object scenarios.

[mlzoo]

Hi! @mlzoo is your code working? I found some syntax errors and many more errors, mainly: "local variable ... referenced before assignment"

Hi can you give me the error information?

[Skwarson96]

@mlzoo At the beginning in line 67: print('Training size:', len() ^^^^^ SyntaxError: Generator expression must be parenthesized

[mlzoo]

@mlzoo At the beginning in line 67:

print('Training size:', len()

                        ^^^^^

SyntaxError: Generator expression must be parenthesized

My fault! Please delete that line... I will fix the bug later.

[Skwarson96]

@mlzoo of course, but despite this, there are still errors, for example line 70 in persam_f_multi_obj.py: training_size = int(os.listdir(os.path.join(images_path, obj_name))) * args.training_percentage) I thing it should be: training_size = int(len(os.listdir(os.path.join(images_path, obj_name))) * args.training_percentage) But when I fix it I still get a lot of errors. Lot of variables are referenced before assignment