Artificial Intelligence Section

The Artificial Intelligence Section is engaged in research and development of technologies for forensic examination using image engineering and information engineering. In particular, our research domains cover the following wide scope of topics:

Forensic image and video analysis
Big data analytics with criminal databases
Fundamental techniques related with artificial intelligence including deep learning

Within the areas described above, we also deliver expert evidence as expert technical witnesses.

Our research themes related with forensic image and video analyses are:

Estimation of physical quantity such as human height and vehicle velocity
Camera identification using CCD fingerprint method
Image enhancement
Person identification using gait analysis
Character recognition for captured objects in images (eg. license-plate numbers and characters printed on clothes)
Detection of tampered images

Image analysis software
An example of image restoration using image analysis software developed in our laboratory.
(a) Before (b) After (c) Result of multi-frame super resolution

Person identification based on gait analysis
An example of gait-analysis-based individual identification from surveillance footage.

Real-time anomaly detection system
Omni-directional object detection for unmanned aerial vehicles (UAVs) using deep learning

Our research themes related with recently developed AI technologies are:

Anomaly detection and suspicious object/person detection with deep learning techniques
Next-place prediction and identification of offender's residential locations for criminal investigation
Shooting-location identification based on image contents
Numerical simulation for accident (incident) reconstruction
Development of systems to support security activities and criminal investigation

R & D on Forensic Image and Video Analyses

R & D using recently developed AI technologies