Tomato Plant Detection

Project overview

Agricultural field images were annotated for a computer vision dataset focused on detecting tomato plants in real-world agricultural environments. Visible tomato plants were identified and accurate bounding box annotations created to generate high-quality training data for an AI-based object detection model.

The main objective was to prepare a reliable dataset for automatically detecting tomato plants across different field conditions and growth stages. As specified by the project, annotations used the Tree class label.

Work performed

• Reviewed and organised the agricultural image dataset before annotation.
• Visible tomato plants were identified in each image.
• Accurate bounding boxes were created around individual tomato plants.
• Labelling remained consistent throughout using the Tree class specified for the project.
• Partially visible plants were annotated where applicable.
• Tomato plants were carefully distinguished from surrounding weeds, crops, soil and other vegetation.
• Plants at different sizes and growth stages were handled throughout.
• Quality checks were performed, and inaccurate, missing or poorly positioned annotations were corrected.
• Annotations were confirmed consistent and suitable for object detection model training.

What made it hard

• Detecting tomato plants in dense agricultural environments.
• Tomato plants had to be differentiated from weeds and surrounding vegetation.
• Overlapping plants and foliage complicated boundary placement.
• Partially visible plants needed accurate annotation.
• Different growth stages produced variations in plant size and appearance.
• Complex field backgrounds made individual plants difficult to identify.
• Varying lighting conditions and shadows affected plant visibility.
• Tight and consistent bounding boxes around irregular plant structures required careful attention to detail.

Result

A high-quality tomato plant detection dataset was completed with accurate and consistent bounding box annotations. The dataset was prepared for training a computer vision model capable of automatically detecting tomato plants in real-world agricultural environments.

Final deliverable

Accurate and consistent bounding box annotations were delivered under the Tree class specified for the project throughout the tomato plant dataset. The final output was prepared according to project requirements and was ready for AI model training and automated agricultural plant detection applications.