Detection of Traffic Sign using Inception V3 in Comparison with VGG-19 to Measure Accuracy

To find traffic signs using Inception V3 and compare the performance VGG-19 algorithm for the detection of traffic signs. The sample size is estimated using G-power software, a total of 44,028 samples are acquired by fixing the parameters such as power and error rate as 0.80 and 0.05 respectively. The traffic sign detection consists of 44,028 sample images of different traffic signs used for training 38,302 (80%) and testing 5,726 (20%) the predictive model in python and the statistical analysis is performed in SPSS. The inception V3 algorithm is used and compared with the VGG-19 algorithm. The predictive model using the Inception V3 results in a higher accuracy of 94.38% than the VGG-19 based model with an accuracy of 90.12% and with a significance value of p=0.123 (p<0.05). The results show that the Inception V3 based model provides significantly better results than the VGG-19 based model in the detection of traffic signs.