Blueberry Plant Segmentation

Project overview

Agricultural images were annotated for a computer vision dataset focused on segmenting blueberry plants in field environments. The work involved creating precise polygon annotations around visible blueberry plants to generate high-quality training data for an AI-based segmentation model.

The annotations were designed to help the model identify blueberry plants accurately while separating them from surrounding vegetation and complex agricultural backgrounds.

Work performed

• Reviewed and organised the blueberry plant image dataset before annotation began.
• Identified visible blueberry plants in agricultural field images.
• Created precise polygon segmentation masks following the visible boundaries of each plant.
• Maintained consistent labelling using the Blueberry_Plant class.
• Partially visible plants were annotated wherever applicable.
• Blueberry plants were distinguished carefully from surrounding weeds, grass, soil and other vegetation.
• Overlapping branches and leaves were handled with detailed polygon annotations.
• Quality checks were carried out and inaccurate or incomplete masks were corrected.
• Annotations were confirmed as consistent and suitable for computer vision model training.

What made it hard

• Blueberry plants had complex structures with many branches and leaves.
• Dense foliage made individual plant boundaries difficult to identify.
• Overlapping branches and neighbouring plants required careful segmentation.
• Similar-looking vegetation and weeds created additional identification difficulties.
• Partially visible plants needed precise boundary estimation.
• Different plant sizes and growth stages demanded consistent annotation standards.
• Shadows, lighting variations and complex field backgrounds affected plant visibility.
• Accurate polygon boundaries around irregular plant structures required significant attention to detail.

Result

A high-quality blueberry plant segmentation dataset was completed with precise polygon annotations. The dataset was prepared for training a computer vision model capable of detecting and segmenting blueberry plants automatically in real-world agricultural environments.

Final deliverable

Accurate and consistent polygon segmentation masks for blueberry plants were delivered across the dataset. The final output met the project requirements and was ready for AI model training and agricultural computer vision applications.