Solanum Nigrum Weed Detection

Project overview

Agricultural field images were annotated for a computer vision dataset focused on detecting Solanum nigrum weeds in real-world field environments. The work involved identifying visible Solanum nigrum plants and creating accurate bounding box annotations 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 Solanum nigrum while distinguishing it from crops, grass and other surrounding vegetation.

Work performed

• Reviewed and organised the agricultural field image dataset before annotation.
• Identified visible Solanum nigrum plants in each image.
• Created accurate bounding boxes around individual target weeds.
• Labelling was kept consistent under the Solanum_nigrum class.
• Partially visible weeds were annotated where applicable.
• Solanum nigrum was carefully differentiated from crops, grass and other weeds.
• Handled plants at different growth stages and sizes.
• Performed quality checks and corrected inaccurate, missing or poorly positioned annotations.
• Ensured the annotations were consistent and suitable for object detection model training.

What made it hard

• Differentiating Solanum nigrum from visually similar weeds and surrounding vegetation.
• Identifying small plants within dense agricultural fields.
• Handling partially visible or overlapping weeds.
• Growth stage differences produced variation in plant size and appearance.
• Complex backgrounds of crops, soil, grass and other weeds made detection harder.
• Varying lighting conditions and shadows affected plant visibility.
• Keeping bounding boxes tight and accurate around irregular plant structures demanded careful attention to detail.

Result

A high-quality Solanum nigrum weed 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 Solanum nigrum weeds in real-world agricultural environments.

Final deliverable

Accurate and consistent bounding box annotations for the Solanum_nigrum class were delivered across the dataset. The final output was prepared to the project requirements and was ready for AI model training and automated weed detection and agricultural monitoring applications.