Parking Space Occupancy Classification via Oblique Camera Vision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the occupancy of parking spaces are inefficient, especially when vehicles are parked non-conformingly, such as at an angle or too far from sensors.
Innovation Solution
A method using at least one camera to capture images of the parking space at an oblique angle, which are then processed by a parking space classifier. This classifier extracts structural features from the images and classifies the occupancy using a machine learning model, allowing for accurate identification of unoccupied or occupied spaces without additional equipment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ultrasonic sensors are installed near each parking space to determine occupancy, then occupancy detection is possible, but the system fails to identify non-conforming parking scenarios such as vehicles parked at an angle or too far from the sensor
Solution Approach 1:
The patent replaces ultrasonic sensors with a camera-based vision system that captures images of parking spaces. This optical substitution allows the system to detect occupancy through image analysis rather than acoustic reflection, enabling accurate identification of vehicles regardless of their position or orientation within the parking space.
Solution Approach 2:
The patent introduces a machine learning model as an intermediary between the camera images and occupancy classification. This intermediary processes images to extract features and determine occupancy status, bridging the gap between raw visual data and meaningful occupancy information, thereby handling diverse parking scenarios effectively.
2Measurement precision
If video surveillance cameras with human staff are used to identify parking space occupancy, then occupancy can be determined, but the system is labor-intensive and inefficient
Solution Approach 1:
The patent implements an automated system where the camera and machine learning model perform occupancy detection without human intervention. The system processes images autonomously, extracts relevant features, and classifies parking space occupancy automatically, eliminating the need for manual monitoring while maintaining high accuracy.
Solution Approach 2:
The patent replaces the manual human monitoring system with an automated computer vision system. This substitution transforms a labor-intensive process into an efficient automated operation, where algorithms process images and determine occupancy status without requiring human staff time or effort.
3Productivity
If simple entry and exit counting is used to estimate free parking spaces, then traffic coordination is possible, but precise occupancy information for individual parking spaces is not available
Solution Approach 1:
The patent divides the parking facility into individual parking spaces, each monitored independently by cameras. This segmentation allows the system to provide detailed occupancy information for each specific parking space rather than just aggregate counts, enabling precise guidance while maintaining overall traffic coordination capabilities.
Data Source
Figure 1~2
Figure 3~4c
Figure 5
AI summary
The invention relates to a method for classifying an occupancy of a parking space as well as a system for classifying an occupancy of a parking space. According to the invention, at least one camera (30) captures at least one image (72) of the parking space (20), in particular at an oblique angle, the at least one image (72) is fed to and processed by a parking space classifier (80), wherein the parking space classifier (80) extracts features from the images (72) of the parking space (20) and classifies the occupancy (92) of the parking space (20) as unoccupied or occupied, wherein the parking space classifier (80) comprises at least one machine learning model configured, in particular trained, to extract features from the at least one image (72) of the parking space (20) and to classify the occupancy (92) of the parking space (20) based on the extracted features.