Abstract:

Abstract Most public areas have parking areas to make it easier for customers to park their cars. With advances and developments in technology, parking lots can smartly solve the problem of finding a parking space that is still empty or cannot be filled. To find out which parking lots are empty or already filled, an algorithm or method is needed and with the help of Computer Vision techniques. In this study, some data were collected and used several types of trained models to find the most accurate results. The results of the research are in two scenarios, namely the parking lots are still empty and those have been filled them. During with empty and filled scenarios where the parking space is empty or has been filled with cars conditions. Based on experiments using the Convolutional neural network (CNN) model, an average percentage of accuracy, precision, recall, and F1 Score are above 90%. Meanwhile, there are several pre-training models in occupied parking lot scenarios such as VGG19, Densenet121, Resnet V2 50 and EfficientNetB7 which can be used for development purposes due to their excellent accuracy. The research results contribute to a better understanding of selecting accurate pre-trained model results for use in development and turnkey applications.

Keywords: