A Novel Framework Using Bald Wolf Optimized DCNN Classifier for Traffic Sign and Obstacle Detection

Traffic sign detection plays a crucial role in smart driving systems. By accurately identifying and interpreting traffic signs, it helps drivers adhere to regulations and significantly decreases the likelihood of accidents. Current traffic sign and object detection techniques work well under typical daylight conditions. However, their effectiveness diminishes in challenging driving scenarios, including low light, changes in illumination, complex road environments, and nighttime conditions. Moreover, existing models overlook the significance of lower-layer features in traffic sign detection, leading to less accurate results. To overcome these challenges, this research introduces an innovative approach for effective traffic sign prediction utilizing the Bald Wolf optimized Deep Convolutional Neural Network (DCNN) Classifier. To effectively detect traffic signs and obstacles, the extracted features are then supplied into the Bald Wolf Optimized DCNN Classifier. The Convolutional Neural Network (CNN) can generate important and individual feature representations for specific data. The DCNN classifier’s analytical capability is leveraged, and efficiency is further Boosted by Bald Wolf Optimization (BWO), optimizing parameters like weights and bias for enhanced convergence. This optimization method combines Bald Eagle and Grey Wolf algorithms, integrating intelligent social and hunting behaviors to fine-tune parameters. For the road sign detection dataset, the results are calculated using the proposed BWO-DCNN, where specificity, sensitivity, and accuracy attain 94.80%, 94.28%, and 94.91%. Similarly, the proposed model attains 94.83%, 93.38%, and 94.03%, respectively, for the Udacity self-driving car dataset, which is more proficient than further techniques.