A neural network has been developed in the Urals to help officials process documents. Specialists from the region's artificial intelligence development center spent several months training it to process scanned files and transfer information into text format.
The neural network used the "convolutional kernel" method: the image was read as a whole, not in fragments, just as a human eye perceives a picture. The developers are pleased with the result: the neural network recognizes the image in 95% of cases.
Routine functions are shifted to artificial intelligence, freeing up specialists' time to solve other tasks, increasing the efficiency of work processes by an average of 5-7 times
For example, the neural network can process dozens of documents in one second, facilitating the work of employees of the Department of Property Relations of the Tyumen Region.
It should be noted that the Ural Center for AI Development appeared on the basis of the Institute of Radio Electronics and Information Technologies. Its specialists create solutions that are in demand in business and government.
Read also:
In Russia, a neural network has been trained to recognize dangerous objects in images