Contrast-Guided Cross-Modal Distillation for Thermal Object Detection
This addresses robust perception at night for thermal detection, offering a mono-modality solution that improves accuracy without extra sensors, though it is incremental in refining existing distillation methods.
The paper tackled the problem of low contrast and weak high-frequency cues in thermal-infrared object detection, which cause duplicate boxes and missed small objects, by introducing training-only objectives that sharpen decision boundaries and inject cross-modal semantic priors, achieving state-of-the-art performance.
Robust perception at night remains challenging for thermal-infrared detection: low contrast and weak high-frequency cues lead to duplicate, overlapping boxes, missed small objects, and class confusion. Prior remedies either translate TIR to RGB and hope pixel fidelity transfers to detection -- making performance fragile to color or structure artifacts -- or fuse RGB and TIR at test time, which requires extra sensors, precise calibration, and higher runtime cost. Both lines can help in favorable conditions, but do not directly shape the thermal representation used by the detector. We keep mono-modality inference and tackle the root causes during training. Specifically, we introduce training-only objectives that sharpen instance-level decision boundaries by pulling together features of the same class and pushing apart those of different classes -- suppressing duplicate and confusing detections -- and that inject cross-modal semantic priors by aligning the student's multi-level pyramid features with an RGB-trained teacher, thereby strengthening texture-poor thermal features without visible input at test time. In experiments, our method outperformed prior approaches and achieved state-of-the-art performance.