Automated Driver Assistance Systems

Developed a real-time video synthesis application on the concept of Automated Driver Assistance Systems for extracting textual information of objects of interest from real-time videos. A CNN model was trained over a custom dataset after evaluating CIFAR-10 and CIFAR-100, for classifying the objects of interest. Also implemented Tesseract OCR for extracting text from the footage. The accuracy achieved during the training and validation phase was around 94% and 89% respectively.