In Japan, the number of traffic accidents caused by elderly drivers is increasing. The National Police Agency of Japan requires elderly drivers over the age of 75 to　have a cognitive impairment test to check their ability of driving aptitude. However, it is difficult to prevent the occurrence of traffic accidents among the elderly by these measures alone. It is necessary for each elderly driver to understand his or her own driving skills and to take appropriate measures.

In this study, a recurrence plot was created from driving behavior data, and a deep learning image recognition technique was applied to detect abnormal driving.