Object Detection in Infrared Images with Different Spectra

Object detection algorithms have made significant advancements in various fields like autonomous driving, medical imaging, surveillance, retail analytics, and industrial automation. This is mainly because they’re transforming how we utilize technology, enhancing performance and safety. In military applications, object detection algorithms can be used with special infrared cameras to help spot dangers accurately, which is crucial for making smart decisions and achieving mission success.In this study, we investigate the performance of the popular deep learning-based object detection algorithms YOLO versions 8 and 9 across various infrared spectra, including near-infrared (NIR), short-wavelength infrared (SWIR), mid-wavelength infrared (MWIR), and long-wavelength infrared (LWIR), with particular emphasis on tank recognition. Our study contributes to understanding how object detection algorithms perform in infrared (IR) images under different spectra, providing insights for the design and optimization of IR imaging systems, particularly in the context of infrared tank imagery.