Crnn Fa Printed 96 Long
An OCR model optimized for Persian language, based on CNN+LSTM architecture, designed specifically for printed/scanned documents
Downloads 2,886
Release Time : 11/27/2023
Model Overview
This model is an upgraded version of the CRNN model for Persian OCR, improving recognition accuracy through expanded training datasets, adjusted input dimensions, and support for more characters.
Model Features
Enhanced Data Support
Training dataset expanded by 5 times with added support for numbers and special characters
Optimized Input/Output
Input image size adjusted to 32x384, maximum output length increased to 96 characters
Mixed Character Handling
Automatically processes numbers and other left-to-right characters mixed in text
Scenario Adaptability
Although optimized for printed text, it can be fine-tuned for scenarios like license plate recognition or handwritten text
Model Capabilities
Persian text recognition
Printed document OCR
Numeric and special character recognition
Mixed-direction text processing
Use Cases
Document Digitization
Persian Document Scanning
Convert printed Persian documents into editable text
Performs best on texts within 50 characters
Automated Processing
Form Recognition
Automatically identify Persian text and numeric information in forms
Featured Recommended AI Models