Effectiveness of Focal Loss for Traffic Sign Detection Using Deep Neural Networks
Deepika, Sharda Vashisth, Prabha Sharma · International Journal of convergence in healthcare · 2023
Recent developments in autonomous driving have created a great demand for precise and computationallyeffective traffic sign-detecting systems. By assisting drivers and assuring their safety, such technology canlessen traffic accidents and fatalities. However, to make such a system deployable, a few crucial accuracyand processing performance problems must be resolved. Real-time performance is sometimes regarded as amust for such an application. RetinaNet, a focal loss-based single-stage object detector, is employed to strikea compromise between accuracy and processing speed concerning the most cutting-edge object detectors. Thedetector is suitable for traffic sign identification since it was developed to overcome the class imbalance problemthat the single-stage detector had. (TSD). The efficiency of the detector was evaluated by combining featureextractors like ResNet-50 and ResNet-101 on two openly available TSD benchmark datasets. Various metricslike memory allocation, mean average precision (mAP), running time, amount of floating-point operations, andmodel parameters are taken into consideration. Evaluation of the detector on several datasets is required toexamine the variance in the performance, and the RetinaNet model is the fastest and best model in terms ofmemory usage, making it the ideal option for the deployment of mobile and affordable embedded devices.