Answers are grounded in this paper and cite their sources.
Motorcycle rider face detection remains a challenging task in intelligent transportation monitoring due to partial facial occlusion caused by helmets, visors, face masks, and variations in image orientation. This study aims to experimentally compare the performance of Region-based Convolutional Neural Network (R-CNN) and Faster R-CNN in detecting and classifying motorcycle rider facial regions. A dataset of 2,000 motorcycle rider images was collected and divided into training, validation, and testing subsets using a ratio of 70:10:20. Rotation-based augmentation at +10° and −10° increased the training dataset to 4,200 images. Both detection architectures were evaluated using learning rates of 0.01 and 0.001 with 80 training epochs. The experimental results showed that Faster R-CNN achieved a higher mean Average Precision at an IoU threshold of 0.50 (mAP50) of 0.8467 compared with 0.7824 for R-CNN at a learning rate of 0.01. Furthermore, Faster R-CNN achieved an accuracy of 89.42%, precision of 96.35%, recall of 91.56%, and F1-score of 93.89%, outperforming R-CNN, which obtained an accuracy of 80.17%, precision of 85.66%, recall of 84.15%, and F1-score of 84.90%. These findings indicate that Faster R-CNN provides better detection and classification performance than conventional R-CNN under the evaluated experimental conditions. The results provide empirical evidence supporting the application of Faster R-CNN for motorcycle rider face detection in intelligent transportation monitoring systems.
No takes yet. Share an insight, caveat, or question.
Muhaqiqin et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: