Safety is a major priority on jobsites, but many hazards can go unnoticed. Machine learning can enhance safety monitoring through pictures, videos and audio recordings taken on the site.
Smartvid.io, a software company based in Cambridge, Mass., hosted a webinar on how machine learning on jobsites can reduce risk and improve safety.
“A ton of videos and photos are captured on projects every day. There are 50 gigabytes of data for a typical project. Most of it ends up unused, or scattered across different systems and devices,” says Josh Kanner, founder and CEO of Smartvid.io. “Our software lets field teams use our mobile app, or integrate with other systems like BIM 360, so that the information is aggregated in one place.”
The software uses speech and image recognition to apply smart tags to potential hazards and compliance issues.
“The tagging engine will bring the safety issues to the attention of the appropriate person so that it can be fixed. All users in the company can have access to the data,” says Kanner.
Skanska used the software for safety tracking at its 2112 Pennsylvania Avenue project in Washington, D.C.
“The smart tags were created by the team out in the field so that they apply to our specific project,” says Oliver Smith, VDC director at Skanska USA.
People can create alternate tags so that when the program hears “floor one” or “first floor” in a video, it understands both to mean the same thing.
“The program allows users to audio annotate what’s happening in the field. It’s a great way for the team to interact with safety. They collaborate and communicate more clearly than before,” says Smith.
Kanner emphasized that the machine learning software is not meant to replace anyone’s job.
“The program elevates the process. It drives behavior and reduces risk. You control what you do with the data and how to manage it internally,” he says.
“We’re not relying strictly on the machine learning. The human factor isn’t being taken out of it,” says Smith.
The program can also be used to track the progress of a project through photos and videos. It uses the audio from a video and uses keywords from the conversation to add smart tags about the progress.