Pea Segmentation

Project overview

Pea images were annotated for a computer vision dataset focused on detecting and segmenting individual peas. The project involved creating precise polygon annotations around peas to generate high-quality training data for an AI-based segmentation model.

Work performed

• Reviewed and organised the image dataset before annotation.
• Individual peas were identified in each image.
• Created precise polygon annotations around every visible pea.
• Consistent labelling was maintained using the pea class.
• Polygons followed the boundaries of individual peas closely.
• Quality checks identified and corrected inaccurate annotations.
• Prepared the annotated dataset for AI model training and segmentation tasks.

What made it hard

• Annotating small peas with precise boundaries.
• Overlapping and closely positioned peas were difficult to separate.
• Round objects made accurate polygon placement demanding.
• Varied lighting conditions, shadows and image backgrounds affected visibility.
• Keeping individual peas properly separated during annotation required care.

Result

A high-quality pea segmentation dataset was completed with precise polygon annotations for individual peas. The dataset was prepared for training a computer vision model able to detect and segment peas automatically.

Final deliverable

Accurate polygon annotations for individual peas were delivered with consistent labelling and high annotation quality across the dataset. The final output met the project requirements and was ready for AI model training and development.