Construction projects are complex undertakings, having an extra set of eyes can decrease the possibility of something being overlooked. The larger the project, the more likely it is that something might get missed. The same set of challenges applies to overall job site safety. While a human inspector could easily spot machinery being driven unsafely or a potential trip hazard on a residential construction project, spotting every risk in every room of a commercial office tower in progress would be a vastly different undertaking.
Artificial Intelligence has already proven itself a valuable technological ally in construction. This runs the gamut, from smart technology ensuring efficient use of construction equipment to advanced data analytics that can unearth actionable insights from huge amounts of project data. As AI’s capabilities have improved, so has the range of what it can be trained to do.
Companies have begun specializing in AI that can track project progress and job site safety by analyzing pictures or video and making independent assessments. This could have big implications for how construction firms monitor risk in the future.
Smartvid.io’s ‘VINNIE’ Beats Human Inspector in Photo Contest Case Study
Procore integration partner Smartvid.io uses a mix of machine learning and image recognition technology to scan job site photos and video footage for potential safety hazards. Using its proprietary AI known as VINNIE (Very Intelligent Neural Network for Insight & Evaluation), Smartvid.io analyzes project images and other job site data, automatically “tagging” instances of what VINNIE is told to observe. This could be as fine-tuned or general as desired; the AI could either tag any photos in which people appear or focus on the details, searching for instances where people aren’t wearing hard hats or visibility clothing.
A few years ago, Smartvid.io partnered with Engineering News Record (ENR) for a case study to test VINNIE’s capabilities. ENR holds an annual Year in Construction Photo Contest, for which readers from C&E firms submit thousands of project images. These images are judged not only on their composition and lighting but are also reviewed by a safety expert for possible safety hazards. For its 2016 competition, VINNIE scanned the very same images alongside the human safety inspector, 1080 in all. VINNIE completed its work in under 10 minutes, while the safety expert took 4.5 hours to finish. VINNIE also found dozens more instances of missing hard hats or safety clothing in its scan.
Perhaps best of all, any images or observations uploaded through Procore are kept in one location through Smartvid.io, which can be easily referenced through a keyword search. The system even produces risk-scored project reports so that companies know where to focus their attention on limiting risk.
Buildots’ Virtual Inspector Sees Uptake in UK
British-Israeli startup Buildots uses AI to capture, upload, and scan job site images captured from GoPro cameras worn on workers’ hard hats. As a supervisor tours a project site, captured video and images are automatically uploaded to the image recognition software. Buildots’ system then compares the images against a digital replica of the finished project, which looks at tens of thousands of objects, down to the finest details like individual bathroom fixtures. According to the MIT Technology Review, the AI can track approximately 150,000 objects multiple times a week, identifying the status of each of them, ranging from “not yet begun” to “fully installed.”
Site inspection is a time-consuming process involving lots of repetition and requiring meticulous and unwavering attention to detail. A system like Buildots makes project progress tracking a significantly more streamlined process, freeing up decision-makers for more important tasks.
“That’s the job people want to be doing—not having to go and check if the walls have been painted or if someone’s drilled too many holes in the ceiling,” Buildots’ Sophie Morris told MIT Technology Review.
Buildots’ technology has already gotten some attention, with two of the biggest C&E firms in Europe already using it, including for several residential building projects by UK construction giant Wates.
AI job site inspection isn’t a technology intended to replace human workers. Instead, it’s a tool to help entire companies better account for and mitigate job site risk. At the same time, it’s ensuring projects aren’t held up because of a snag that could have been caught and addressed earlier. Until there’s a way to be in multiple places at once, the all-seeing eye of artificial intelligence is as close as we can get.