Key points are not available for this paper at this time.
Traffic sign detection is a pivotal aspect of intellect transportation systems, which play a crucial role in enhancing road safety along with traffic management. The utilization of the YOLOv5 framework provides a holistic solution for determining independently as well as recognizing road symbols. This approach was assessed using the German Traffic Sign Recognition Benchmark (GTSRB), a comprehensive data set comprising over 50,001 images across 43 distinct traffic sign classes. The model achieved an impressive testing accuracy of 95.31%. This abstract investigates the efficacy of the Pelican Optimization Algorithm (POA) and Cuckoo Search Algorithm (CSA) in fine-tuning hyperparameters for traffic sign detection. POA draws inspiration from the collaborative hunting behavior of pelicans, employing a swarm intelligence approach to adjust parameters such as learning rates, network architecture, and augmentation techniques. Likewise, CSA, inspired by the brood parasitism of cuckoo birds, utilizes Levy flights and random walks to effectively explore complex parameter spaces, finding an equilibrium among exploration and extraction. identify optimal model configurations. Both algorithms demonstrate proficiency in optimizing critical parameters essential for traffic sign detection, encompassing network architecture, training parameters, and data augmentation strategies. They adept exploration mechanisms and collaborative methodologies contribute to bolstering the accuracy and reliability of traffic symbol identification systems. In conclusion, the strategic utilization of metaheuristic optimization algorithms like POA and CSA significantly enhances the performance of traffic symbol identification by adeptly fine-tuning hyperparameters. These algorithms serve as valuable tools in optimizing detection models for real-world applications, with opportunities for further refinement and exploration of hybrid methodologies to maximize efficacy.
Kumaravel et al. (Fri,) studied this question.