Aerial Parcel Image Classification for Automated Land Improvement Valuation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for determining the value of real estate with human-made land improvements are costly and time-consuming, requiring manual visits to parcels of land.
Innovation Solution
A neural network is trained using localized training data to classify and localize human-made land improvements in digital aerial images, generating a report that includes the improvements, their values, and the parcel's value, leveraging convolutional neural networks (CNNs) and U-Net for image classification and segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual visits to parcels of land are used to determine real estate value, then valuation accuracy is improved, but cost and time consumption increase
Solution Approach 1:
The patent uses aerial images as copies of the actual land parcels to perform valuation assessments. Instead of physically visiting each parcel, the system creates and analyzes digital copies (aerial photographs) to identify improvements and determine values, thereby eliminating time-consuming manual visits while maintaining assessment accuracy
Solution Approach 2:
The patent replaces the mechanical system of manual field visits with an automated image processing system. Convolutional neural networks and computer vision algorithms substitute for human assessors physically traveling to parcels, automatically detecting and classifying land improvements from aerial images to generate valuation reports
2Measurement precision
If manual visits to parcels of land are used to determine real estate value, then valuation accuracy is improved, but cost increases
Solution Approach 1:
The system uses aerial image copies to assess land values, eliminating the need for expensive manual field visits. The digital copies allow multiple assessments to be performed on the same data without additional travel or fieldwork costs
Solution Approach 2:
The system enables self-service valuation where the automated image processing and neural network algorithms independently perform the assessment without requiring human field assessors. The technology serves itself by automatically detecting, classifying, and valuing improvements from aerial images
3Productivity
If automated image classification is used to detect land improvements, then productivity is improved, but measurement precision may worsen
Solution Approach 1:
The system performs preliminary training of convolutional neural networks using labeled datasets of aerial images before deployment. This preliminary action ensures the automated system learns accurate patterns for detecting and classifying different types of land improvements, maintaining precision while enabling high-speed automated processing
Solution Approach 2:
The system uses feedback loops where neural networks are continuously trained and refined using ground truth data from manual assessments. The automated detection results are compared against known accurate values, and the system learns from discrepancies to improve future detection accuracy while maintaining high productivity
Data Source
AI summary
A method for identification of land improvements in a given parcel of land that includes generating a first training data set by clipping a large image of a parcel of land into individual parcel images each including images of improvements, and each improvement being labeled with an improvement type, providing the individual parcel images to a first classification model, training the first classification model based on the individual parcel images to identify unlabeled improvements in a parcel image and to obtain a multi-label classifier, generating a second training data set of images, and training a second, semantic segmentation model based on the second training data set.