Pea Detection

Project overview

The Pea Detection project involved annotating agricultural field images to identify pea plants for computer vision model training. The dataset was prepared to support accurate pea detection in real-world agricultural environments.

Work performed

• Reviewed agricultural field images and identified pea plants.
• Accurate bounding box annotations were created around the plants.
• Peas were differentiated from surrounding crops, weeds, soil and vegetation.
• Annotated plants of differing sizes and levels of visibility.
• Overlapping and partially visible plants were handled individually.
• Consistent annotation quality was maintained across the dataset.
• Missing or inaccurate annotations were reviewed and corrected.
• Prepared the dataset for pea detection model training.

What made it hard

• Distinguishing pea plants from surrounding vegetation and weeds.
• Small and partially visible plants were difficult to detect.
• Overlapping plants in dense field areas complicated separation.
• Complex agricultural backgrounds reduced plant visibility.
• Keeping bounding boxes accurate and consistent across varied image conditions took careful attention.

Result

A high-quality annotated dataset was prepared for pea detection, suitable for training computer vision models for agricultural crop detection and monitoring.

Final deliverable

The final dataset contained accurate and consistent bounding box annotations for the pea class and was prepared for computer vision model training and agricultural monitoring applications.