Voiz: Automated Traffic Sign and Obstacle Detection Assistance using Tensorflow and CNN

The tremendous progress of society and the economy has made autos one of the most practical means of transportation for practically every family today. As a result, the road traffic environment is becoming more complex, necessitating the development of an intelligent voice-assisted application to govern driving operations based on data from traffic signs and the detection of real-time barriers. Vehicle cameras may collect photos of the road in real life, allowing for detection, identification of roadside hazards for the traffic signs. The driving system receives precise data thanks to the alerts the system issues to the driver in the form of simulated voice commands for the various conditions observed. In order to process the deep features inside the picture autonomously based on the training samples for the target identification, this project makes use of the deep learning open source library tensorflow. Tensorflow is a computer vision technique which helps us in detecting and tracing an object. The special attribute about TensorFlow is that it identifies the class of the obstacle (person, pothole, animal, object) and their location-specific coordinates. The trained model focuses on alerting the driver prior to the location on the basis of the traveling speed. The GTTS (google-text-to-speech) engine is used for the audio alert system. It takes text and turns them into spoken language. The estimated accuracy of the above system proves to be efficient as compared to the standard systems available by means of detection and assistance.