FIELD: computer equipment.
SUBSTANCE: invention relates to means of recognizing symbols using neural networks. Method comprises: assigning an input image of grapheme to a first cluster of grapheme from a plurality of clusters of grapheme, where the first grapheme cluster comprises a first plurality of grapheme, selection by a processing device of a classifier from a plurality of classifiers of a second level, a grapheme classifier based on a first cluster of grapheme, such that the selected classifier has been trained to recognize the first plurality of grapheme, and processing the input image of the grapheme using the selected classifier to recognize at least one symbol on the output image of the grapheme. Systems ensure operation of the method.
EFFECT: efficient and accurate character recognition for languages with a large alphabet and multilingual texts.
20 cl, 9 dwg
| Title | Year | Author | Number |
|---|---|---|---|
| DIFFERENTIAL CLASSIFICATION WITH MULTIPLE NEURAL NETWORKS | 2017 |
|
RU2652461C1 |
| HANDWRITING RECOGNITION USING NEURAL NETWORKS | 2020 |
|
RU2757713C1 |
| NEURAL NETWORK TRAINING BY MEANS OF SPECIALIZED LOSS FUNCTIONS | 2018 |
|
RU2707147C1 |
| OPTICAL CHARACTER RECOGNITION USING SPECIALIZED CONFIDENCE FUNCTIONS, IMPLEMENTED ON THE BASIS OF NEURAL NETWORKS | 2018 |
|
RU2703270C1 |
| METHOD OF DETECTING NECESSITY OF STANDARD LEARNING FOR VERIFICATION OF RECOGNIZED TEXT | 2014 |
|
RU2641225C2 |
| AUTOMATIC DETERMINATION OF SET OF CATEGORIES FOR DOCUMENT CLASSIFICATION | 2018 |
|
RU2701995C2 |
| OPTICAL CHARACTER RECOGNITION BY MEANS OF COMBINATION OF NEURAL NETWORK MODELS | 2020 |
|
RU2768211C1 |
| IDENTIFICATION OF WRITING SYSTEMS USED IN DOCUMENTS | 2021 |
|
RU2792743C1 |
| RECOGNITION OF EVENTS ON PHOTOGRAPHS WITH AUTOMATIC SELECTION OF ALBUMS | 2020 |
|
RU2742602C1 |
| TEXT RECOGNITION USING ARTIFICIAL INTELLIGENCE | 2017 |
|
RU2691214C1 |
