Toshiba AI Technology Catalog

  • Media recognition
  • Sensor data recognition

Visual Relocalizer

Estimate camera’s position from photo based on background captured.

  • This technology estimates the mobile camera’s position and pose by comparing to a database generated form camera images obtained in advance.
  • The database is generated automatically from data measured using camera and LiDAR. Even without LiDAR, it can be formed from camera images by using 3D reconstruction technologies.
  • Using deep learning in database referencing and camera movement estimation, this technology is able to accurately estimate positions.

© OpenStreetMap contributors


  • Autonomous movement of vehicles, drones, and robots
  • Automation of maintenance and inspection services

Benchmarks, strengths, and track record

  • Able to estimate positions from images alone, with an error of several 1 cm (indoors) to several 10 cm (outdoors).
  • No need to install visual markers.


Contact the Toshiba Corporate Research & Development Center

Please include the title “Toshiba AI Technology Catalog: Visual Relocalizer” 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.


  • R. Nakashima and A. Seki, “SIR-Net: Scene Independent End-to-End Trainable Visual Relocalizer,” 3DV, 2019.