Luminaire Detection through Brightness Enhanced Mask R-CNN in Tunnel Environment

Abstract. Rapid luminaire detection enables effective remote monitoring and management, thereby facilitating intelligent tunnel lighting maintenance. Despite its powerful object detection capabilities, deep learning methodologies encounter challenges in tunnel luminaire detection due to the complex environment and unfavorable lighting conditions. To overcome these issues, this paper proposes an improved tunnel luminaire detection solution by enhancing the Mask R-CNN using brightness balancing. Leveraging tunnel gray-scale images and the Mask R-CNN object detection framework, a feature fusion network based on ResNet-FPN, trained via transfer learning, which enhances performance in detecting object luminaires. Furthermore, considering the differences in luminaire brightness and their backgrounds, an object brightness enhancement method based on Kapur’s Entropy Method is introduced to effectively reducing missed detections and false positives, thereby improving the detection rate of luminaires. To evaluate the performance of the proposed approach, real datasets of tunnel environment are used. Experimental results revealed that the proposed approach achieved precision, recall, an F1-score and AP50 of 94.9%, 82.3%, 0.881 and 0.776, respectively, which improved of 4.3%, 4.4%, 0.044, and 0.151, respectively, compared to the original model, thus, could be applied to the 3D model construction and intelligent management of tunnels.