datumaro

datumaro copied to clipboard

Currently, our implementation of RISE algorithm works only with OpenVINO plugin and requires a custom model output parsing script from user. The algorithm works, but it is hard to use....

zhiltsov-max

Support 3D data

1

Vision datasets often include 3d data - it can be lidar, MRI, landscape, buildings and models. It would be nice is Datumaro could handle such datasets. Examples: BRATS, KITTI, ObjectNet3D,...

zhiltsov-max

zhiltsov-max

Add dataset quality checks

1

When there is a dataset, it is often important to be able to check its quality. There are few way to check it: - compare annotations against a specification. Find...

zhiltsov-max

converting voc->coco for segmentation task

5

I've converted the voc data to coco format for segmentation task and tried to train OTE instance segmentation models with the data. ``` voc.export(out_path, "coco", save_images=True, reindex=True) ``` However, OTE...

jihyeonyi

I've tried to `get_bbox` of Mask with an image below. ``` Mask(image = [ [ 0, 0, 0, 0, 0 ], [ 0, 0, 1, 1, 1 ], [ 0,...

jihyeonyi

CVAT export loses data

1

I use CVAT video 1.1 annotation format, and I want to change the label name. I modified the example from the documentation, but I lose some attributes like: - track...

illiaSkif

### Summary - Added ability to parse masks with holes into polygons - Added support of holes in polygons TBD: - [ ] Update polygon/mask use in formats ### How...

zhiltsov-max

Extend existing documentation with format implementation guidelines, cover such topics as: - Requirements - Error handling - Allowed deviations from the spec / original files - Task-specific dataset representations and...

zhiltsov-max

Requirements: - more dedicated support of track annotations - ReID annotations - images as annotations (image-to-image tasks - super resolution, style transfer, image generation, etc.) - (depth-)field annotations (distance prediction...

zhiltsov-max