KupynOrest/instance_augmentation

This project helps computer vision researchers and practitioners expand their image datasets by redrawing individual objects within scenes, keeping their original shape and context. You input your existing images with object annotations (like bounding boxes or segmentation masks), and it outputs new, diverse images where specific objects have been replaced with variations while the rest of the scene remains consistent. This is ideal for those training models for object detection, instance segmentation, or saliency detection.

No commits in the last 6 months.