Detection of Vehicle and Pedestrian in Indian Challenging Scenarios

Intelligent visual perception technology is crucial for providing accurate vehicles and pedestrian to ensuring road safety in autonomous vehicles is crucial, as the escalating rate of road accidents. However, complex road environments can cause information on vehicles and pedestrians to become unclear, and changing light angles can cause the image to become darkened, which makes detection harder. In his paper, four separate YOLOv9 models are trained separately to improve the robustness of the detection system in varied weather conditions. During testing, samples are classified using the SVM classifier and pass the classified into trained YOLOv9 model for specific environments. The results demonstrate YOLOv9 stands out with higher mAP@50 scores of 96.60%, 94.30%, 95.90% and 82.10% for morning, afternoon, evening and night samples respectively for Indian Dataset. This study contributes insightful information about the development of weather-aware vehicles and pedestrian detection systems, offering a comprehensive analysis of the model's performance.