ryouchinsa / Rectlabel-support

RectLabel is an offline image annotation tool for object detection and segmentation.
https://rectlabel.com
504 stars 73 forks source link

Export mask images #223

Closed chaewonS closed 1 year ago

chaewonS commented 1 year ago

I tried opening the color mask image using the code below. originalFormat = np.array(Image.open(f))

And I checked that the pixel value of the image is indexed by the class id. if it is a person, all of them are output as 25, but I want to distinguish each person (for example, 25000, 25001, 25002) Is there any solution?

ryouchinsa commented 1 year ago

Thanks for writing the issue.

For the semantic segmentation, the indexed color mask image is exported, and each pixel value is defined by the object id. If there are multiple person objects in the image, all pixel values in those persons have the same pixel value defined by the object id. For the instance segmentation, use "Export an image per object" option. If there are multiple persons in the image, for each person, each grayscale mask image is exported.

If you have any questions, please let us know.

chaewonS commented 1 year ago

Thanks for the answer! Where can I use "Export an image per object" option?

ryouchinsa commented 1 year ago

You can choose on the Export -> Export mask images menu.

スクリーンショット 2023-05-17 15 59 28