- Media recognition
Inspection information management technology for infrastructure maintenance
Reduce labor in inspection tasks by recognizing and automatically managing inspection information from a single image.
- Recognizes the capture location and size of the object with a degradation (e.g., crack) from an inspection image taken with a regular camera.
- Inspection information recognized from the image is centrally managed in a database and linked to positional information on a drawing.
Applications
- Plant facilities; e.g., power plants and oil refineries
- Large-scale factories and warehouses
Benchmarks, strengths, and track record
- To estimate positions in indoor facilities, competitors need additional equipment such as wireless base stations and markers, but Toshiba’s method can estimate the capture position in 3D from a single image.
- To measure the size of an image object, competitors need additional equipment such as laser scanners or depth sensors, but Toshiba’s method can estimate the object size from a single image.
Inquiries
Contact the Toshiba Corporate Research & Development Center
Please include the title “Toshiba AI Technology Catalog: Inspection information management technology for infrastructure maintenance inspections” 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:
- R. Nakashima and A. Seki, “SIR-Net: Scene Independent End-to-End Trainable Visual Relocalizer,” International Conference on 3D Vision (3DV), (2019)
- Kashiwagi, N. Mishima, T. Kozakaya, and S. Hiura, “Deep Depth From Aberration Map,” IEEE International Conference on Computer Vision, (2019)