Stick Recognition

Project overview

Agricultural field images were annotated for a computer vision dataset focused on recognising sticks and wooden objects present in field environments. Visible sticks were identified and enclosed in accurate annotations to generate high-quality training data for an AI-based object detection model.

The annotations were designed to help the model distinguish sticks from crops, soil, weeds and other objects commonly found in agricultural fields.

Work performed

• Reviewed and organised the agricultural field image dataset before annotation.
• Identified visible sticks within the field environments.
• Drew accurate bounding boxes around individual sticks.
• Applied the Stick class consistently across the dataset.
• Annotated partially visible sticks where applicable.
• Distinguished sticks carefully from crop stems, branches, weeds, soil and other field objects.
• Performed quality checks and corrected inaccurate or incomplete annotations.
• Ensured the annotations were consistent and suitable for computer vision model training.

What made it hard

• Detecting thin and small sticks within large field images.
• Crop stems, branches and dry vegetation were easily mistaken for sticks.
• Partially hidden and overlapping sticks complicated placement.
• Sticks lay at many different angles and orientations.
• Complex backgrounds contained soil, crops, weeds and natural debris.
• Lighting conditions and shadows sometimes made sticks difficult to identify.
• Keeping bounding boxes accurate around thin, irregularly positioned objects required care.

Result

A high-quality stick recognition dataset was completed with accurate and consistent annotations. The dataset was prepared for training a computer vision model capable of automatically recognising sticks in agricultural field environments.

Final deliverable

Accurate and consistent annotations for visible sticks were delivered across the dataset. The final output was prepared to the project requirements and was ready for AI model training and agricultural field monitoring applications.