- Media recognition
- Anomaly detection
Model-based image anomaly detection
Detect specified anomalies (e.g., cracks, rust, or road damage) from images.
- Using a deep learning model , detect the presence of specified anomalies (e.g., cracks, rust, or road damage), and the location of those anomalies in images.
- A model can be trained to detect anomalies down to their location in images that are annotated only for the presence or absence of anomalies on an image-by-image basis, making it easy to perform additional training with local data.
Applications
- Labor reduction through automation of infrastructure inspections such as tunnel cracks and road paving
Benchmarks, strengths, and track record
- Verification testing currently underway with Central Nippon Expressway Company.
Inquiries
Contact the Toshiba Corporate Research & Development Center
Please include the title “Toshiba AI Technology Catalog: Model-based image anomaly detection” or the URL in the inquiry text.
Please note that because this technology is currently the subject of R&D activities, immediate responses to inquiries may not be possible.
References:
- Satoshi Itoh, "Multiple type road surface anomaly detection using Multiple Instance Learning," Workshop on practical applications for ViEW Vision technologies; OS1-O3, Dec. 2022 (in Japanese).
- Toshiba and Toshiba Digital Solutions Develop AI for Road Surface Anomaly Detection that Contributes to Maintenance and Long-term Operation of Expressways and Demonstrates
Real-time Detection of Potholes that Pose a Risk of Serious Accidents
-Promoting more rapid and sophisticated routine inspection of expressways using AI to facilitate detection of various anomalies and easy implementation on different roads- – R&D Center; Toshiba (global.toshiba) - Shirakawa et al, “Automating Daily Inspection for Expressways Using Anomaly Detection Model, ” PHMAP23, 2023.9.
