Fig. 4: Evaluation on real-world driving.
From: Embodied neuromorphic synergy for lighting-robust machine vision to see in extreme bright
a Acquisition system. b Exposure time. c Gain. d Over-exposed rate. e Mean average precision (mAP). f Object detection results when the acquisition vehicle is driving out of a tunnel. The proposed NEC effectively manages the substantial increase in natural lighting when the vehicle exits the tunnel, ensuring consistent detection by swiftly adapting to this drastic change in lighting conditions and preventing the occurrence of over-exposed images. This avoids any potential compromise in the accuracy of the detection process (see t = 137.5 s in f).
