Identification and identification of passport data

Identification of necessary text blocks, the recognition of objects, the identification of text.
In the input we have an image, and in the output JSON file with data. The image can be returned, have any background - all this works. The model takes into account any distortions of the type of perspective and others.