A UAV-imagery dataset for cotton ball detection and counting on cotton crops. The dataset contains 180 images with 5,333 annotated cotton ball instances (single class: cotton_ball). Bounding boxes are in COCO format [x, y, width, height].
This augmented version extends the original CBDA collection with additional synthetic samples to improve model robustness across varying altitudes, lighting conditions, and crop… See the full description on the dataset page:
https://huggingface.co/datasets/Project-AgML/CBDA_Cotton_Ball_Detection_Augmented.
