There is an incredible amount of traffic signs on the road. Although they are fairly passive, they still get a lot of wear and tear. Maintenance workers have to keep track of them, in order to know their condition. This is a terribly annoying and sometimes even a dangerous job. However, scientists have just developed AI tool, which is able to keep track of street sign locations by tapping into Google Street View images.
Manual geo-location of traffic infrastructure is a hugely important and yet strenuous job. Furthermore, because workers have to wonder close to active traffic, it can also be quite dangerous. And it is still necessary, because street signs wear and traffic situations change. This new technology detects signs with near 96% accuracy, identifies their type with near 98% accuracy and can record their precise geo location from the 2D images. The best thing is that it can use such open databases as the Google Street View, accurately marking the location of each signs.
Scientists trained the system to recognize STOP and GIVE WAY signs, but other types of objects can be plugged into the algorithm too. Dr Chayn Sun, co-lead author of the study, said: “This imagery is critical for local governments in monitoring and managing assets and with the huge amount of geospatial applications flourishing, this information will only become more valuable. Ours is one of several early applications for this to meet a specific industry need but a whole lot more will emerge in coming years”.
Images can be interpreted and catalogued to identify signs that are in need of immediate attention. This will allow replacing damaged signs quicker and preventing unnecessary dangerous jobs of the workers. Furthermore, this AI system can use other sources of imagery, not just Google Street View. For example, videos recorded by garbage trucks or buses can be used too. This is especially useful because Google Street View images are often several years old. This technology could really improve the management of the assets of local governments, but it can also be used by business institutions in industrial setting just as well.
The system still needs some refinement, but it is already much more effective that currently used methods. Scientists took a look at existing databases and found that GPS location data in existing street sign databases was often inaccurate, sometimes up to 10m off. In many cases this could become a problem. And this is where the AI system has an advantage.
Source: RMIT University