Transform

Here we provide dataset transformation examples such as reidentification, reindexing, label redefinition, and tiling.

In addition to these, Datumaro provides a total of 22 transformations, including polygon to bbox, merging segmentation masks, annotation and attribution removal, etc. Please refer here.

Transform Dataset: Re-id, Reindexing, Remapping, etc.

Tile your Dataset to Cope with High-Resolution Images

This transform is known to be useful for detecting small objects in high-resolution input images [1].

Transform bounding box to instance mask annotations using Segment Anything Model

This transform uses Segment Anything Model [2] to transform bounding box annotations to instance mask annotations.

Automatic instance mask annotations generation using Segment Anything Model

This transform uses Segment Anything Model [2] to generate instance maks annotations automatically.

References

[1]

F., Ozge Unel, Burak O. Ozkalayci, and Cevahir Cigla. “The power of tiling for small object detection.” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. 2019.

[2] (1,2)

Kirillov, Alexander, et al. “Segment anything.” arXiv preprint arXiv:2304.02643 (2023).