Abnormal driving behavior detection via Deep Learning Approaches
Tanusree Roy, Ancy Omanakuttan · International journal of advance research, ideas and innovations in technology · 2020
During this period detection of abnormal driving is more important. It give safety to passengers and drivers in vehicle . In our proposed method deep learning methods used for abnormal driving behavior prediction. Deep learning classification methods are more applicable of various fields. The applications are computer vision, speech finding, language processing, audio identification, medical image analysis, bioinformatics and self driving cars. Our simplified method we use three deep learning methods such as CNN, Deep Residual Network and Visualized Geometric Group16. The CNN extract high level features such as hidden object in an image and doesn’t need hand designed features. For the purpose we use one dataset that contain 22,424 color frames (images).In the existing system that use novel deep learning fusion methods to detect abnormal driving behavior and the accuracy of that dataset is 82%. Our proposed method revealed an accuracy 87.44%. So new method more better than the existing method.