Video analytics with generative AI: a new paradigm

Video analytics is one of the richest sources of information available to organizations. However, with the solutions available before the advent of generative AI—which were based on rigid rules and object detection—, their full potential was not fully leveraged. Opportunities to take real-time action, prevent incidents, or retrospectively analyze critical events were often missed.

Generative AI is here to transform this paradigm: it enables contextual understanding to interpret and make sense of complex environments while introducing temporal awareness to track events over time. Moreover, thanks to virtual assistants that interact in natural language, it democratizes access to complex video data analysis, allowing users to simply request the information they need and receive instant results. No technical expertise is required.

On one hand, it automatically detects critical events, such as equipment failure or unauthorized access of a person or vehicle to a specific area, and translates these insights into immediate re-sponses—triggering alerts or executing predefined actions. On the other hand, this "all-seeing assistant" can answer all kinds of questions about the monitored environments. A logistics company user can inquire about what happened at the loading dock in the morning, a security officer can ask how many people passed through a restricted area during a given period, and a nurse can check what time breakfast was served in a specific room.

An important clarification: integrating generative AI into video analytics is not just about replacing old technology with new. It’s about enabling capabilities that were previously impossible. This solu-tion unlocks the untapped potential of video and drives transformative outcomes across all indus-tries.

Looking ahead, expectations are even higher: generative AI agents are expected to work in sync to fully automate the entire process: from incident detection to resolution. For instance, if a video ana-lytics agent detects a fire in a facility, it can interact with another agent that will autonomously call the fire department without human intervention.

The first step has already been taken: a significant number of companies use cameras in their op-erations—factories, hospitals, logistics centers. It’s time to process those images and turn them into actionable insights for better decision-making.

NTT DATA is positioned to bridge generative AI technologies with the real needs of its clients, set-ting new frontiers for intelligent monitoring.

— Carlos Porto Filho, Data Science Manager at NTT DATA