Cotton Detection & Size Classification

Project overview

Cotton plant images were annotated for a computer vision dataset focused on detecting cotton bolls and distinguishing between regular and small cotton. The work involved creating accurate bounding box annotations around individual cotton bolls and assigning the appropriate class based on their size.

The dataset was prepared as training data for an AI model capable of automatically detecting cotton and identifying small cotton bolls.

Work performed

• Reviewed and organised the cotton image dataset before annotation.
• Identified individual cotton bolls on the plants.
• Created accurate bounding boxes around visible cotton.
• Each annotated cotton boll was classified as Cotton or Small Cotton.
• Consistent class labelling was maintained across the dataset.
• Partially visible cotton was annotated where applicable.
• Performed quality checks and corrected inaccurate bounding boxes or class labels.
• Ensured annotations were suitable for object detection model training.

What made it hard

• Detecting cotton bolls among dense leaves and branches.
• Distinguishing between Cotton and Small Cotton on the basis of visible size.
• Handling overlapping and partially occluded cotton bolls.
• Small cotton had to be annotated accurately in complex backgrounds.
• Maintaining consistent bounding box placement and class assignment.

Result

A high-quality cotton detection dataset containing the Cotton and Small Cotton classes was completed. Each visible cotton boll was assigned the appropriate class and enclosed with an accurate bounding box, and the dataset was prepared for training an AI model capable of detecting cotton and distinguishing small cotton in agricultural images.

Final deliverable

Accurately annotated cotton images were delivered with bounding boxes for the Cotton and Small Cotton classes, with consistent labelling and annotation quality maintained throughout the dataset. The final dataset was prepared according to project requirements and was ready for AI model training and development.