Construction sites are equipping themselves with video recognition technologies
AI is revolutionizing the world of industry, and video recognition technologies in particular are already proving their worth in many sectors with automatic checkouts in stores or quality control, but it would seem that they are becoming more and more important in other areas that are less cited in terms of innovation, yet just as promising, such as construction.
Far from the clichés of a lagging sector, the construction industry is following the trend and adopting video recognition on construction sites in order to closely monitor their operations and develop their safety measures.
In concrete terms, how could a video recognition device be used on a construction site? All you need to do is place a few cameras around the site and equip yourself with a panoramic camera to get a global view of the space. These cameras would be equipped with a video recognition system trained to recognize humans, machines, and equipment. This system would make it possible to extract valuable information for the daily management of the site and its follow-up. Equipment manufacturer Komatsu, for example, has chosen video recognition to monitor its construction sites with the help of NVIDIA technology. They have set up a 3D visualization of construction sites with real-time monitoring of the interaction of people, machines and connected objects in order to improve site security.
The first advantage of such a system is the increased safety of people on the site. Indeed, despite a drop in the rate of workplace accidents on construction sites, this sector still has a significant and recurring number of accidents: 88,273 accidents in 2016, including 112 deaths. Despite the progress, the sector remains one of the most affected by workplace accidents. Manual handling operations and falls are still the most frequent causes of accidents at work. The mandatory wearing of personal protective equipment (PPE) is therefore an essential issue which must be closely monitored by construction companies in order to protect their workers.
Video recognition can help them, with the installation of cameras capable of detecting both when workers fail to wear PPE and when there is dangerous proximity between a human being and a machine. This is what the machine manufacturer Colas, in partnership with Volvo CR, has done. They have equipped construction sites in Switzerland with a person detection system, with integrated AI, to assist the machine operator in their work by alerting them when a person is detected near their machine. In addition, detection algorithms are able to notify the site manager in real time if a worker does not wear his equipment.
The second advantage is the protection of the site itself. The already installed cameras can also act as intelligent video surveillance cameras by detecting intrusions and notifying the relevant authorities in real time of theft or damage to materials or equipment. When we know that almost all French departments are affected, and construction companies of all sizes are targeted by criminal acts, combined with the fact that the cost of these acts of vandalism on construction sites is estimated by the FFB to be at least 1% of the turnover of the construction sector, i.e. more than one billion euros, we can easily see the financial gains that such an intelligent video surveillance system represents.
Finally, these various applications combined can be used to measure the “health” of a site, its smooth running and its long-term progress. Indeed, through the detection of individuals and machines, it is possible to know the attendance of the site and quantify its activity, its movement, the number of machines on site etc.. For operational staff, these indicators make it possible to monitor and understand directly the dynamism of each site and increase its productivity.
However, let us conclude these fine promises with a cautionary note. Two conditions are essential for the adoption of AI on construction sites in order to derive maximum benefit from it. First, we must guarantee the anonymity of employees. The information collected by the smart cameras must be used for the protection and overall monitoring of the site, and not for tracking a particular individual. Then, it is necessary to scale management by bringing together the different beneficiaries: safety and security, environmental hygiene, production management; and to guarantee a system able to distribute the information to the different beneficiaries.
It is under these conditions that the construction industry will be able to really exploit and benefit from the resources offered by video recognition. This collaboration between industry and technology is also an additional opportunity for AI actors to prove that the technology is viable on an industrial scale.