From Buzzword To Reality: How AI Is Powering Video Technology To Augment Human Performance
By Dr. Barry Norton, VP of Research, Milestone Systems
It seems like Artificial Intelligence (AI) dominates the news headlines in ways that few other technologies have managed to. What threat, if any, does its ‘intelligence’ present? How will it interact with human beings?
Rather than serve as a replacement, we believe AI will augment people in the workplace. Human-led insight and guidance will remain essential and critical, but collecting enough data on human intent is challenging. Businesses will increasingly turn to synthetic data to provide this training and to mitigate problems of scale, privacy and bias. Synthetic data is ‘artificial data’ containing computer-generated information detached from real-world data, such as images or video. This protects the privacy of individuals, while still ensuring accuracy, as the data is generated in a way that maintains the characteristics of real-world data without containing actual personal information.
Recently, Singapore announced its Proposed Guide on Synthetic Data Generation, for businesses to make sense of synthetic data. It is timely in providing best practices and clarity in creating synthetic data. The G7 AI Code of Conduct is a strong guideline for advanced AI systems. To build more comprehensive video analytics systems, the industry must recognise the fundamental role of appropriate data input measures and protections for personal data and intellectual property.
Video pioneers the use of AI
At present, much of the excitement around AI concerns its potential in application across modalities — e.g. text, images, video and audio — for instance in vision-language models (VLMs) such as OpenAI’s Sora. This could potentially extend to safety and security applications. In fact, there is one overlooked industry that has already made technological leaps — embedding AI and even synthetic data — video technology. Once used solely for security-focused applications, video is now used to build new safety-centric environments for everyday citizens.
Basic video analytics include fundamental functions like object detection, recognition and tracking, These functions are relied on for security applications internationally. For example, in spatial domains, object detection has been used for counting people, protecting perimeters, and identifying when objects cross defined lines. In the so-called ‘temporal’ domain, object tracking is used to extract information relating to the trajectory of objects — for example, to assess the direction of moving cars in traffic.
At the next level, second-level analytics addresses the interpretation of those objects and their behaviours across frames, making action recognition, interaction detection and anomaly detection all possible. Importantly, these analytics do not have to be focused on an object alone: so-called ‘reconstruction-based’ anomaly detection can in fact work with an entire frame of video. This makes it possible for use in life-saving applications, like detecting when people fall.
Putting synthetic data to work
Synthetic data was successfully leveraged for real application in a research project with the University of Aalborg, Denmark. This synthetic dataset supplemented the primary dataset —thermal images — to detect when an individual has fallen into the Aalborg Harbor. The primary challenge was getting the AI video surveillance system to accurately identify falls.
First, the researchers had volunteers carry out simulations of these falls —jumping and diving into the water —to provide the system with data represented by different motion patterns. Then when it became impractical to accurately recreate unintended falls into the water, the researchers enhanced the training data for the system with synthetic data. This synthetic data simulated scenarios involving more intricate behaviour patterns, such as fights, falls from wheelchairs, skating, and biking. This addition made the AI video surveillance system more comprehensive. The Aalborg Harbor is now a safer place for walking and activities.
Overcoming challenges in AI advancement
Future developments in data-driven video analytics are intrinsically linked to the development of AI. Advanced video analytics relies on large, annotated datasets on which to be trained, with corresponding rights for ethical use. Ready-to-use datasets are in short supply, while the cost of creating and labelling new datasets can be considerable.
This is why synthetic data is useful and practical. Artificially generated or augmented, synthetic data provides a means of increasing the amount of data available on which to train an AI model. This in turn reduces the need for manual annotation and extensive data collection and delivers training data that more fully represents the diversity of human experience. Importantly, synthetic data can preserve real-world characteristics and safeguard privacy, while also avoiding consent-related issues.
Unlocking the potential of AI-driven video
The continued development of AI on which data-driven video technology depends very much relies on the wider acceptance of AI by consumers, businesses and governments. For video surveillance, which has its roots in the use of CCTV cameras for security applications, that trust has already been established. Now that AI underpins innovative applications for data-driven video, it remains important that the technology is used in responsible ways.
As AI is developed further, the evolution of video technology will draw on work in related fields being undertaken today and into the future. As such, the video sector should perhaps be seen as a ‘sandbox’ that delivers real-world applications for AI, providing users of data-driven video technology with value beyond security.
