Handwritten Paragraph Text Recognition using OCR

The recognition of unlimited signatures is an ongoing challenge for computer vision systems. Traditionally, in paragraph text recognition, two separate models are used for character splitting and text character recognition. In this regard, this study proposes an integrated end-to-end framework and a new hybrid thinking approach to address the existing challenge. The prototype is specifically intended to handle a splitting picture per align iteratively, with three different modules. Initially, the encoder creates maps of features derived from the whole passage picture. Then, a concept module produces typically a mask with weights, which enables focused analysis of current text line features, especially implicit line splitting Finally, the decoder module is used for performing spectrum detection in each text line feature, ending with comprehensive detection of the entire paragraph offers a new solution.