Download Full Text (1.2 MB)
Matthew Stranz, CIS494: Research in Computer Information Systems
Faculty Mentor(s): Professor Sarbani Banerjee, Computer Information Systems
The purpose of this research project is to create a vehicle dash camera that assists the user, police, and insurance company in easy license plate recognition and number retrieval in an event of an accident and/or hit and run. Presently, personal vehicle dash cameras are very limited and only record data. Since accidents are very chaotic, it can be difficult extracting the required information. To implement these specific aspects, the vehicle license plate will be detected by YOLOv4, a real-time object recognition system and machine learning model. Following detection, the number within the license plate will be filtered through OpenCV real-time computer vision and printed with Tesseract OCR optical character recognition engine. Currently, this is a software capability that is only found in government, police, drone, and stationary security cameras. Supporting software for this research will be with Google Collaboratory, Jupyter Notebook, Anaconda (Miniconda for Raspberry Pi), Nvidia CUDA Toolkit 10.1, and Git. Additionally, the hardware that would be needed is a Nvidia GPU, Raspberry Pi 4 Model B, a storage device micro SD card, Arducam day and night vision camera, Raspberry Pi 7" touch screen display, USB to USB Type-C with 15W car charger, and a housing unit that would incorporate the hardware along with a mounting for the windshield or the mirror. The final goal is to have all license plate numbers be enlarged and printed above the license plate, and all detections be saved onto the storage device.
Stranz, Matthew, "Vehicle Dash Cameras with Artificial Intelligence" (2021). Computer Information Systems and Engineering Technology. 5.