Predictive Analysis of Vehicle Accident using GoogleNet Classifier Compared with ResNet-50

This research aims to make real-time predictions about road traffic accidents using two distinct machine learning (ML) techniques and to compare the performance of selected classifiers. The study suggests utilizing neural networks, specifically the GoogleNet and ResNet-50 classifiers, to achieve this goal.GoogleNet algorithm is used in comparison with ResNet-50 with different 80% of training and 20% of testing data in order to predict vehicle accidents. The average G-power over the course of the tests is approximately 95%. The dataset for the present work contains 12,300 unique values with group 1=9840 and group 2=2460 related to the light condition, weather conditions, types of collisions and the factors considered are reviews. On performing statistical analysis (SPSS) on group 1 and group 2 and achieved mean accuracy of 94.3110% and 87.33% respectively. It expresses that there is a statistical significance variance between the Novel GoogleNet progression and the ResNet-50 algorithm with p = 0.001 (p<0.05). According to the testing outcomes, the proposed approach is believed to be the best technique for identifying accidents that happen on the road as a result of vehicle collisions. The investigation’s efficacy is evidenced by the higher level of accuracy demonstrated by the proposed Novel GoogleNet classifier when compared to both the ResNet-50 classifier and traditional classifiers typically employed in accident prediction.