Specialists from the Moscow Technical University of Communications and Informatics (MTUCI) are working on a system that can read damaged barcodes. Traditional methods of restoring them have their limitations, so researchers are looking for new solutions.
The system is based on neural image restoration algorithms. When scanning a code, the user transmits data to a neural network, which determines the location of the code and restores the lost elements.
Recognition accuracy depends on the image quality: the minimum resolution should be at least 200 DPI, and two-dimensional codes require about five pixels per smallest element.
MTUCI scientists will test the model on data with varying degrees of damage, and then test it on real-world tasks. Their work in the future will help avoid errors in inventory management in warehouses and medications — in hospitals.
Read more on the topic:
MTUCI explained how lightning affects the sensitivity of optical fiber
Scientists in Russia are training a neural network to detect fake voices
Neural network for creating films and TV series developed in Russia
