The data from Google’s Street View cars is being screened through an advanced new algorithm developed by Google and reCAPTCHA. The new algorithm is designed to filter the data in order to identify difficult to read objects such as street numbers.
This algorithm is ground breaking as digitally deciphering a street number from a picture taken from a moving car has always been problematic due to factors such as blur, poor lighting and shadows. Obtaining exact street numbers is a critical part of making the Street View data match with Google Maps when calculating the exact position of a house. It is also a key aid in providing directions.
Google Inc. (NASDAQ:GOOG) and reCAPTCHA released the joint paper at the International Conference on Learning Representations (ICLR). The paper also points to other applications of this advanced algorithm such as it can be used to read CAPTCHAs. CAPTCHAs have been an irritating security feature on many websites for a long time forcing a user to input a code that is bent out of shape or hidden in a crowded background.
This could pose a potential problem for sites using only this type of human verification as this makes it clear that it is no longer only humans that can decipher CAPTCHAs. For Google’s reCAPTCHA, this is not a great concern as they have been reducing their reliance on this type of verification for some time now. Instead, they have been depending on more advanced risk analysis to determine if the user trying to enter is human or a machine. For the everyday user, it could mark the end of squinting to find a code to enter a website.
