ConvertAnnotationsToDagsHub/YOLO

Converting YOLO Annotation to DagsHub Format

The YOLO (You Only Look Once) format is a specific format for annotating object bounding boxes in images for object detection tasks. In this format, each image in the dataset should have a corresponding text file with the same name as the image, containing the bounding box annotations for that image

YOLO Dataset

For this tutorial the "Traffic Signs" dataset was chosen.

NOTE: The notebook requires that all images be stored in the images folder and all the text files (with the exception of classes.txt) be stored in the labels folder. The dataset was downloaded and manually grouped into folders and then uploaded to the repo. To make it easier, this repository contains the dataset in a pre-grpuped format.

Implementation

This folder has the following stucture:

YOLO
|   Convert_YOLO_Annotations_to_DagsHub_Format.ipynb
|   README.md
|   
\---data
    |   classes.txt
    |   
    +---images
    |       
    \---labels

The images are all stored under the images folder and the YOLO annotations nder the labels folder.

NOTE: It is required that you push the entire data folder to your repo prior to running the colab notebook.

To convert YOLO Annotations to Dagshub Format:

• Create a new DagsHub Repo
• Follow the instructions in the colab notebook

VOILA!

DEMO

We tested the transfer annotations too. Checkout these repsotories:

-Traffic-Sign-Classifications -Traffic-Sign-Classification -Private-Transfer-Annotations-DVC - This one's on a sqirrel dataset.