DEEP CONVOLUTIONAL NEURAL NETWORK FOR PARKING SPACE DETECTION AND CLASSIFICATION
Car parking is a major problem in developing countries, to overcome this problem we go for intelligence parking system. This work affords and compares specific imaginative and prescient-based total approaches to estimate the occupancy popularity of parking regions by using counting cars and non-empty parking stalls. Our research considers for both the scenarios, wherein parking stalls are marked at the ground and where in no assumption at the presence or position of stalls is believed. We perform an experimental analysis on a real-international dataset of motion pictures occurred in unique parking areas. Techniques like image classification and object detection using Mask RCNN are carried out in different scenarios with different architectures. Our analysis highlights that: (1) techniques based on image classification with different architectures can be correctly leveraged (2) techniques based on object detection with Mask RCNN need to be preferred over strategies based on image classification.