I'm very new to using YOLO and this is one of my assesments where i have to train a model using YOLOv7. I have stated my query below and also any guidance would be appreciated
https://www.kaggle.com/datasets/hassanmojab/xview-dataset
can someone take a look at the above datsset and let me know what format is required for the labels when using YOLOv7 for object detection and how to convert .geojson to that particular file format.
You need one txt file per image.
One line in the txt file per object.
Line format:
class x-center y-center width height
No headers. No commas. Classes start at 0. Coordinates with respect to the width/height of the image (i.e. always between zero and one)
Example:
img01.txt
0 0.5 0.5 0.1 0.1
1 0.1 0.1 0.2 0.2
This corresponds to an object of class 0 in the center of the image and an object of class 1 in the top left corner.
In addition, a yaml file with the names of the classes and path to data must be provided in the yolov7/data folder.
As for the particular dataset you have, looks like you have polygons, you have to calculate a bounding box for those.
