To create a dataset for YOLO-based object detection, we compile 1500 images across four classes: buffalo, elephant, rhino, and zebra, preprocessed for optimal model training. Each image is annotated with bounding boxes and class labels in YOLO format, which includes class ID and normalized bounding box coordinates. The dataset is strategically divided into training, validation, and testing sets with an 8:1:1 ratio, respectively, totaling 1200 images for training, 150 for validation, and 150 for testing. This careful preparation and distribution ensure a comprehensive dataset that facilitates effective model learning, validation, and evaluation, which is critical for achieving high performance in object detection tasks.

Now available in Ultralytics - https://docs.ultralytics.com/datasets/detect/african-wildlife/