- YOLOv4 format only works with Image or Video asset type projects that contain bounding box annotations. Exporting other annotation types to YOLOv4 will fail.
- YOLOv5 format only works with Image asset type projects that contain bounding box annotations. Exporting other annotation types to YOLOv5 will fail.
- If you want to export labels created using polygons and semantic object detection, you must use the Kili SDK.
- YOLOv7 and YOLOv8 are currently only available for exports done programmatically. For details, refer to Kili SDK documentation.
classes.txt (YOLOv4) or
data.yaml (YOLOv5) contains IDs and names of all labeling classes used in a job.
images folder contains all the images (or a list of links to images) from the dataset being exported.
labels folder contains a list of text files:
- Each text file contains the list of annotations for the specific image.
- Each annotation is a line in the text file; it contains the annotation class ID and the coordinates of the bounding box.
Updated about 1 month ago