- Media recognition
- Media data analysis
- Media generation
AI-powered metadata generation technology for broadcast data management
We analyze the video and audio of TV programs to automatically generate metadata that represents the content of those programs by leveraging Toshiba’s media AI technology and open-source technology.
- Leveraging Toshiba's image recognition, speech recognition, and acoustic signal processing technologies, along with open-source modules and models, we automatically generate metadata for TV programs.
- This technology generates metadata that includes cut times, scene times, details about people and objects in the video, sound events (like background music and applause), transcriptions, keywords, and summaries.
Applications
- Social systems (ES / Roads / Communications / Broadcasting)
Benchmarks, strengths, and track record
- We are considering the use of this technology in the proposed Broadcast Databank project by Toshiba Infrastructure Systems & Solutions Corporation.
- Demonstration experiments creating POP displays and digest movies from metadata and setting them up in retail stores have confirmed a reduction in POP setup time and a threefold increase in sales.
Inquiries
Contact the Toshiba Corporate Research & Development Center
Please include the title “Toshiba AI Technology Catalog: AI-powered metadata generation technology for broadcast data management” 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:
- In demonstration experiments conducted jointly by Toshiba Infrastructure Systems & Solutions Corporation, Nagoya Broadcasting Network Co., Ltd., and Heiwado Co., Ltd., utilizing broadcast data from information programs for sales promotions, the impact on profit increase at retail shops was confirmed. (In Japanese)
- Face recognition AI Kaometa™
- Toshiba’s Auto-Subtitling System for Online Classes is a Win-Win for Educators and Students
- Kobayashi, D. "Self-supervised Prototype Conditional Few-Shot Object Detection". Image Analysis and Processing ‒ ICIAP 2022: 21st International Conference. Lecce, Italy, 2022-05, Italian Association for Computer Vision, Pattern Recognition and Machine Learning (CVPL- ex-GIRPR). Springer, 2022, p.681‒692.
