The LWIRPOSE project introduces over 2,400 high-quality LWIR thermal images to facilitate human pose estimation under complex real-world conditions, such as lighting changes and occlusions. Critical highlights of this research include:

• The dataset includes detailed 2D human pose annotations, making it a robust resource for researchers and practitioners.
• Captured across various everyday activities, it challenges pose estimation methods to perform accurately despite occlusions.
• Benchmarks of state-of-the-art pose estimation methods on this dataset establish a strong baseline for future work.
• The practical implications for surveillance, healthcare, and sports analytics are profound, offering the potential to significantly enhance these fields.

This comprehensive dataset emboldens the push towards more reliable and versatile pose estimation techniques that can be applied in a multitude of settings. Making such data publicly available catalyzes further collaboration and innovation in the community, paving the way for transformative applications in various industries. Explore the dataset.