ABSTRACT
In both developed and developing countries, ensuring the safety of people with mobility impairments (PRM) on the road is a major concern. These folks are getting a lot of attention right now. Russia, for example, has a profusion of individualized programs to improve their lives and, more especially, to guarantee that they may safely use the transportation infrastructure in numerous cities. Before training the model, there is preprocessing, and then there is feature extraction. To address these issues, the actionable mining model employs two components of the preprocessing step: variable selection and recoding. A set of features is extracted from the video stream during the feature extraction procedure. When training the model, we opted for the ResNet50-LSTM because of its superior accuracy. With an accuracy of 96.08%, the suggested strategy outperforms competitors like ResNet50 and LSTM.
