Open shahaamirbader opened 2 years ago
If I understood your question correctly, you need an object detector, while the code is for segmentation. I think you can apply quad-tree approach for detection too, and sequentially look for objects in the image, with different precisions.
I would like to use this network on an image, thats already been segemented. The image consists of various indoor objects. Is there a way to let the network know, which objects we want the quadtree to be applied on? For example if I have two objects with in a segmented image obtained from an instance segmentation network i.e. a chair and table, is there a way to achieve the following: