Deep Learning is a new growing learning technology that can be used to locate illegally parked vehicles successfully with 99%.
If you live in a city and you intend to buy a car, in addition to road taxes you should also consider the cost of parking. In fact, you should calculate the cost of calls for illegal parking that you will accept at some point.
And that's because as technology develops, automatic road policing measures increase. Last month at congress 2017 International Conference on Deep Learning Technologies a team of researchers from Xidian University in China demonstrated a detection system based on Deep Learning and who can diagnose cars that have parked illegally with 99% accuracy.
The system described at the conference uses what is known as a Single Shot Multibox Detector (SSD). It is a species neuronal network that simplifies object tracking by creating a set of predefined "original boxes". When the scanning starts, the algorithm compares the contents of these boxes to a basic level.
A prediction model is produced from object observations.
Xidian's team's method detects cars and so things about it processing of objects are a little simpler because the cars are more or less the same. This reduces the computational complexity of the processors thus allowing the new system to operate in real time.
Detection of illegal parking does not take place for the first time.
A corresponding experimental program had been implemented in Maryland a few years ago, achieving accuracy rates of about 95 percent. A critical difference between this system and most other methods is that it was based on magnetic sensors rather than optical identification of objects.
The relative report of the system states that this protects the privacy of citizens. Naturally in Greece 99% is not covered by the law, and for the sake of 1% we will never see it on the busy streets of our cities.
