NTT’s Tsuzumi AI learns to read images; Expected to be used for reading foreign maps and menus

SAN FRANCISCO – NTT Corp. has developed an advanced image reading technology to enhance the functionality of its Tsuzumi Generative AI, the company said. Tsuzumi is expected to be fully commercially available by the end of 2024.

Generative AIs have had trouble understanding visual information such as graphs and charts. NTT’s new technology will be able to read articles, websites and contracts that contain graphics or illustrations, and produce responses about or summarize such images.

NTT expects the technology will be used to summarize foreign maps and restaurant menus captured by smartphones, and said its accuracy is comparable to Google LLC’s latest generative AI platform, Gemini. NTT also plans to develop speech and other recognition technologies.

In March, NTT began offering tsuzumi, which has a high level of Japanese language proficiency, to companies and local governments in Japan. NTT said it has received more than 500 inquiries about the AI ​​so far.