Annotated Thermal Image Dataset for Car-Window and Person Segmentation

This dataset, developed under the EU-funded project ODYSSEUS (Grant Agreement ID: 101073910), contains annotated thermal images designed for car, window, and person segmentation tasks. The dataset supports research in computer vision, autonomous surveillance, and thermal scene understanding, and is primarily formatted for YOLO-based object detection and segmentation models.

Each image includes detailed bounding boxes and segmentation masks, enabling accurate model training and evaluation for thermal-based object recognition. The dataset contributes to advancing AI-driven situational awareness and intelligent monitoring systems, as part of ODYSSEUS’s mission to enhance border security and operational efficiency through cutting-edge AI technologies.