Poultry Chick Counting

Project overview

The Poultry Chicks Counting project involved annotating poultry farm images to identify and count individual chicks. The dataset was prepared for computer vision model training focused on automated chick detection and counting.

Work performed

• Reviewed poultry images and identified individual chicks.
• Accurate bounding box annotations were created around each chick.
• Every chick was annotated separately so counting stayed accurate.
• Overlapping and closely positioned birds were handled individually.
• Consistent annotations were maintained across differing image conditions.
• Reviews identified missed or incorrectly labelled chicks.
• Prepared the dataset for automated chick detection and counting.

What made it hard

• Counting chicks accurately when they were positioned very close together.
• Overlapping and partially visible birds were hard to separate.
• Small chicks in crowded areas were easy to miss.
• Bounding boxes had to stay tight without including surrounding background.
• Ensuring no chick was left out during annotation required careful review.

Result

A high-quality annotated dataset was prepared for automated poultry chick detection and counting, suitable for training computer vision models.

Final deliverable

The final dataset contained accurate and consistent bounding box annotations for individual chicks, making it suitable for computer vision model training and automated chick counting applications.