Kawasaki, Japan, December 12, 2024 – Fujitsu Limited has launched a revolutionary video analytics AI agent to improve safety and efficiency in frontline workplaces. The AI derives actionable insights from spatial video and photo data, along with written documentation for improvements in workplace environments.
Selective attention mechanisms applied to multimodal large language models (LLMs) enable the agent to deal with long-duration videos with unmatched accuracy. First attempts at more than two hours of benchmark tests show the best performance worldwide on challenges such as drops in frame rates and poor storage efficiencies.
The AI, an important part of Fujitsu’s “Kozuchi” AI service, will be tested in the newly developed FieldWorkArena supervised by Carnegie Mellon University. It simulates actual operations to give researchers benchmarks and tasks.
Fujitsu plans to launch trial environments in the fiscal year 2024 and start internal implementation by January 2025. Applications will include but are not limited to logistics and construction, automating task updates and visualizing real-time production data.
Fujitsu’s innovation showcases its commitment to improving workplace safety and efficiency by using AI to tackle complicated operational problems.
Source: https://www.fujitsu.com/global/about/resources/news/press-releases/2024/1212-02.html
Latest Stories:
BenevolentAI Reshapes Operations to Lead Next-Gen Drug Discovery
Workday Introduces AI Features to Transform Employee Engagement
STMicroelectronics Unveils First STM32 IoT Module with Qualcomm Collaboration
What is your reaction?
