Artificial vision techniques are evolving fast, as well as their fields of application. For many years, Sacyr Maintenance has been betting on these detection techniques, aimed at continuous state analysis and because of the good results obtained, we continue to work hard to improve current techniques and processes.
With the AI Road project, in collaboration with the tech company Intelygenz, and with the financial support of the CDTI, we are developing a detection and analysis project that will enhance the continuous state analysis and digitization. Artificial vision techniques can help us improve the effectiveness and efficiency of routine surveillance and inspection operations through process automation.
The proposed solution is based on the implementation of an asset recognition and damage classification system using Computer Vision techniques. The footage obtained by a conventional recording system is processed through a detection model that allows locating different assets of interest and classifying them in order to identify damaged ones.
"Thanks to this system, we can detect deteriorated signs, graffiti, damage to beaconing elements, etc. This reinforces surveillance, inspection and inventory operations. To obtain the best results, the most important thing is to generate a good Dataset (sufficient data set) with which to train the models," explains Jorge Zarzuelo, Head of Innovation, Procurement and Machinery Fleet at Sacyr Maintenance.
The end result is a management tool able to self-inventory and to detect missing elements, all located and represented in a simple and straightforward way.