arXiv:2606.01694v1 Announce Type: cross Abstract: Thermal pedestrian MOT remains challenging because weak appearance cues and frequent detection interruptions cause severe trajectory fragmentation. We study whether lightweight post-processing can recover identity continuity without relying on heavy re-identification models or complex online association. Starting from a YOLOv8 and SORT baseline, we add a modular identity-repair backend consisting of online short-gap remapping and offline tracklet relinking based on temporal, spatial, motion, and border cues. Controlled ablations on a fixed vali
Rapid advancements in computer vision and machine learning are pushing the boundaries of identity recognition in challenging environments like thermal imaging, making such post-processing crucial for robust applications.
Improving identity continuity in thermal video enhances the reliability of autonomous systems, surveillance, and security applications, potentially reducing errors and increasing situational awareness.
The ability to recover identity continuity in thermal video with lightweight post-processing means less reliance on computationally intensive methods, making these systems more efficient and deployable.
- · Defence contractors
- · Security sector
- · Autonomous vehicle developers
- · AI software providers
- · Companies relying solely on traditional vision for thermal applications
- · Systems with high trajectory fragmentation
More accurate and persistent object tracking in low-light or obscured conditions becomes feasible.
This could lead to enhanced performance in security systems and autonomous navigation scenarios where thermal imaging is critical.
Broader adoption of thermal imaging in consumer or industrial applications, driven by increased reliability and lower processing overheads.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG