Mobile Camera Colour Drift and Robustness of Lightweight Visual Models under High-Temperature Conditions: A Narrative Review

High-temperature operation affects mobile vision at capture as well as during inference. Sunlight and sustained computation heat the lens-sensor assembly, where changes in focus and dark signal are subsequently modified by the image signal processor. This narrative review examines how hardware athermalisation, image restoration and lightweight visual models address the resulting degradation. Chen et al.'s temperature-conditioned recovery provides direct thermal evidence, whereas cross-camera colour constancy and compact task models contribute methods largely tested under other conditions. The distinction matters when judging transfer to a phone in sustained use. Optical compensation can retain detail that software would otherwise have to estimate; learned correction remains dependent on the camera and its calibration data. A faster task model may still make an unreliable decision from the corrected image. Across the reviewed studies, pairwise integration is better established than a complete thermal pipeline spanning capture, correction and task adaptation. Evaluation needs to distinguish device identity from thermal state and measure image quality alongside task accuracy, energy and latency during real heating and cooling. A lightweight, camera-aware design is plausible, but its cross-device reliability remains to be demonstrated.