Tomato Plant Health Detection

Project overview

Tomato plant images were annotated for a computer vision dataset focused on detecting and classifying tomato plants by their health condition. Individual plants were identified and then separated into healthy and damaged categories.

The main objective was high-quality training data for an AI-based model capable of automatically detecting and differentiating healthy tomato plants from damaged plants in agricultural environments.

Work performed

• Reviewed and organised the tomato plant dataset before annotation.
• Identified individual tomato plants in the agricultural field images.
• Classified each visible plant according to its health condition.
• Created accurate bounding boxes around individual plants.
• Healthy plants were labelled Healthy_Plant and visibly affected plants Damaged_Plant.
• Maintained consistent class labelling throughout the dataset.
• Partially visible plants were annotated wherever applicable.
• Assessed visible signs of plant damage and poor health with care.
• Carried out quality checks and corrected inaccurate, missing or inconsistent annotations.
• Ensured the final dataset suited training a computer vision detection model.

What made it hard

• Differentiating healthy tomato plants from damaged plants by visual characteristics alone.
• Early or less visible signs of plant damage were difficult to spot.
• Partially visible and overlapping plants complicated box placement.
• Growth stages varied, producing differences in plant size and appearance.
• Weeds and surrounding vegetation sometimes obscured plant identification.
• Changing lighting conditions and shadows altered the apparent health of plants.
• Consistent classification standards across a large dataset demanded careful quality control.

Result

A high-quality tomato plant health detection dataset was completed with accurate annotations for the Healthy_Plant and Damaged_Plant classes. The dataset was prepared for training a computer vision model capable of automatically detecting and classifying tomato plants according to their health condition.

Final deliverable

Delivered accurate and consistently labelled bounding box annotations for Healthy_Plant and Damaged_Plant throughout the dataset. The final output matched the project requirements and was ready for AI model training and automated tomato plant health monitoring applications.