Optimasi Sistem Parkir Cerdas Menggunakan Esp32-Cam, Ssd Mobilenet V2 dan Easy Ocr

Abstract

The high number of vehicles in urban areas makes it difficult to find available parking spaces, so drivers often spend time looking for empty slots. This study designs a Smart Parking System that combines Internet of Things (IoT) and Artificial Intelligence (AI) technologies to improve efficiency and accuracy in the automatic parking reservation and validation process. This system uses ESP32-CAM combined with HC-SR04 ultrasonic sensors and RGB LEDs to provide real-time slot avAIlability, then sends the data to the Firebase Realtime Database. To identify vehicles, the system applies the SSD MobileNet V2 algorithm to detect license plates, which are then read using Easy OCR. Image capture is done automatically via ESP32-CAM connected to the server. The user interface is built with React Native so that users can reserve parking slots and connect the status directly through their devices, while the backend uses ExpressJS to ensure smooth data communication. The test results show that the system has high speed and accuracy, with an F1 Score of 97.71%, Precision 96.97%, Recall 98.46%, overall accuracy 96.96%, and OCR accuracy reaching 100%. This system has been proven to be able to speed up the search for parking slots and minimize errors in the vehicle verification process.