Convert Annotations to DagsHub Format

Warning: This repo is using the old git-based flow and is outdated. To learn how to use annotation with datasets click here and check out the annotation converter

YOLO and COCO are two of the most commonly used Dataset formats. DagsHub in collaboration with LabelStudio provide a platfrom to visualize images along with their annotations. But labelstudio does not support YOLO or COCO formats.

To solve this issue, DagsHub provides a functionality to convert YOLO and COCO annotation formats to DagsHub Annotation format that is suppoted by labelstudio and the DagsHub Visualizer.

This Repository provides a methods to achieve the goal of converting COCO and YOLO annotations to the inhouse annotation format.

The Repository has the following tree structure:

├───COCO
│   └───data
│       ├───annotations
│       └───images
└───YOLO
    └───data
        ├───images
        └───labels

Implementation

The Repository has two folders - YOLO and COCO. Each folder contains all the necessary resources to try out the DagsHub annotation conversion functionality.

To try it out:

• Clone the repo
• Zip the data folder under the corresponding annotation folder
• Create a new repo to test the annotations
• Each folder had a custom transfer-annotation colab notebook. Follow the colab notebook
• Click on any image in your repository and you should see the results under the annotation tab.

More deatils are available in the notebooks.