Hi there! Do you still remember our previous blogpost? We elaborated a bit upon the current processes of assessing different situations. We also stated that we have some ideas on how to integrate smart eyewear into those processes. We found out that a big part of artificial intelligence (AI) can help with this proces. In some articles on similar projects, we saw a potential application for AI. Normally, AI refers to ‘making computers think’. A lot of work on AI really focuses on making the computer think like a human. But how can this contribute to our project?
AI actually refers to multiple aspects. One huge part is computer vision. But is it even possible to make a computer see like a human you ask? Computer vision has been around for a while now actually; scanning codes (barcodes, QR-codes, and Snaptags for example) is an everyday example of computer vision. But also near-futuristic applications of computer vision can be seen in self-driving cars for example: these vehicles need to ‘see’ the environment, just like human beings are supposed to do while driving a car. In an article on a ‘traffic safety guardian system’, Chang and Wang implemented computer vision to make driving a car safer: “To decrease car accident rate, we design and implement the traffic safety guardian (TSG) system to assist drivers when some anomaly driving behaviors appear” (288). They built this system using OpenCV and android (288).
This system, however, mainly focuses on other cars and lanes (288). Modern and self-driving cars are able to recognize traffic signs. It is this ‘traffic sign recognition’ (TSR) that can be really useful for our project. First of all, recognizing traffic signs will solve a big part of our problem, for checking the state of traffic signs is one of the many tasks of the analyst. But recognizing objects in general could be really useful in the daily and monthly checks. This is why we started experimenting with OpenCv and android. We managed to build a couple of prototypes which you can see below. We hope that we can further utilize this prototype and the OpenCV library so that we’ll be able to recognize even more objects.
With finishing these prototypes, our current sprint is coming to an end. We managed to get some great insights during this sprint that can be used in the coming sprints!
- Chang, Kuei-Chung, and Kuan-Hsiung Wang. “Design and Implementation of Traffic Safety Guardian System for Android based on OpenCV.” International Conference on Connected Vehicles and Expo (ICCVE) 2012. IEEE, 2012.