Thermal and RGB Image Annotation for AI-Based Emission and Behaviour Monitoring
RoboflowImageBounding BoxPolygon
Labeled and annotated thermal and RGB images of livestock environments to train YOLOv8-based computer vision models for behaviour and emission monitoring. Created bounding box and polygon masks for segmentation of animals and emission areas, ensuring accuracy and consistency across more than 5,000 frames. Applied Roboflow for dataset management, augmentation, and quality assurance. Conducted cross-validation and precision checks, achieving over 97% detection accuracy. Maintained clear documentation and version control to support reproducibility and model improvement.
Labeled and annotated thermal and RGB images of livestock environments to train YOLOv8-based computer vision models for behaviour and emission monitoring. Created bounding box and polygon masks for segmentation of animals and emission areas, ensuring accuracy and consistency across more than 5,000 frames. Applied Roboflow for dataset management, augmentation, and quality assurance. Conducted cross-validation and precision checks, achieving over 97% detection accuracy. Maintained clear documentation and version control to support reproducibility and model improvement.
2024 - 2025