AI and computer vision are here to help work safety
Video surveillance is a common system to ensure facility security in manufacturing, logistics, and many other industries. One of the most interesting and crucial area it can monitor is the implementation of occupational health and safety measures. Regulations and training courses cannot guarantee the proper implementation of the measures or prevent injuries in real time. Companies need a tool not prone to human error. That’s where digital transformation technologies come into play.
Computer vision and AI
Artificial intelligence can monitor what is happening at the sites. Solutions with a surveillance, response, and warning system turn everything the cameras see into valuable information for analysis. It also allows prompt response on emergency situations and access controls for unsafety areas as well.
What can be achieved using AI and computer vision powered safety solutions?
Detection of safety equipment (shoes, reflective vests, respirators, etc.) from the entrance to the work area
Layout of special zones and development of rules for finding in zones
Recognition of employees at the entrance to the work area
Accounting for the time spent in the work area
Integration of control systems with a throughput system
Informing of control services about potential violations
Marking potential violations for subsequent analysis
Cluttering ways of exit of employees (spare and basic)
Out of the box vs. fine-tuned analytics
In general, the video analysis capabilities of such specifically developed systems are vastly superior to that of out-of-the-box solution detectors, which do not always adequately assess the situation, give many false positives, and operate only on a high-quality video stream.
New generation video analytics detectors can be configured, combined, cascaded, and fine-tuned to meet the business requirements. Even low-resolution video streams and analog video surveillance systems can be used to detect events with at least 95+% accuracy. A special tagging toolkit with a built-in self-learning neural network allows companies to quickly tag the video stream and get the content for further training of classifiers—from 1,000 images per hour per one tagger.
Our solution: VizorLabs H&S
Through Softline, many independent software vendors offer their solution focusing on specific aspects of productivity. VizorLabs’s outstanding system is powered by Azure cloud platform that enables a versatile toolkit for creating efficient solutions for specific business tasks. VizorLabs H&S can track whether the employees at the facility wear helmets, gloves, hoods, and safety belts. It also monitors the movement of personnel on roads, the passage of heavy-duty transport, smoking, walking through a restricted-access zone or stairway. Nevertheless it controls the entrance of people into hazardous areas at multiple levels.
The user-friendly interface enables prompt visual control of occupational safety through color-coded icons. Archive footage from any camera can be accessed instantly. Alarm events with date and time are displayed automatically in the event feed. Reports on security violations are also generated automatically for the specified periods.
How does it work from the client’s aspect?
At one of Softline’s clients the video surveillance and analytics system was used to monitor whether the employees wear personal protective equipment. The system controls if they wear helmets, protective shields, and gloves and if their clothes are fully fastened. It also tracks the position of workers relative to switchgear panels. The monitoring system prototype met the customer's expectations, and it was decided to upscale the system.