Automatic license plate recognition (LPR) is indispensable for the admission and flow control of 
vehicles into parking lots orcondominiums.  Generally, these systems are based on classic computer 
vision techniques owing to their processing speed, however, these approaches can lead to inaccurate 
detections and vague performance on non-ideal environmental conditions. My work tried to surpass 
these setbacks with the implementation of an image-based plate recognition system using 
convolutional neural networks (CNN) to enhance the current methods. The system was optimized and 
embedded in a Nvidia Jetson Nano to run in a low-cost computer at a recognition rate of 100ms per 
plate making it ideal to operate in the places mentioned before (Implemented in Python, C++, 
TensorRT/Linux).