Datura Stramonium Detection

Project overview

The Datura Combined project involved annotating agricultural field images to identify Datura stramonium plants. The dataset was prepared to support computer vision model training for accurate detection and monitoring of Datura stramonium in agricultural environments.

Work performed

• Reviewed agricultural field images and identified Datura stramonium plants.
• Created accurate bounding box annotations around the target plants.
• Datura stramonium was differentiated from surrounding crops, weeds, and vegetation.
• Partially visible and differently sized plants were annotated where applicable.
• Maintained consistent annotation quality throughout the dataset.
• Reviewed annotations and corrected inaccurate or missing bounding boxes.
• Prepared the dataset for object detection model training.

What made it hard

• Distinguishing Datura stramonium from similar-looking weeds and vegetation.
• Identifying plants in dense and complex field backgrounds.
• Handling partially visible or overlapping plants.
• Accurate bounding boxes had to be maintained for plants of different sizes and growth stages.
• Ensuring consistent annotations across varying lighting and image conditions.

Result

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

Final deliverable

The final dataset contained accurate and consistent annotations for Datura stramonium and was prepared for computer vision model training and agricultural weed detection applications.