Year: 2024
Type: conference-paper
Abstract: Computer vision tasks include image-based sequence recognition for offline handwriting recognition. Conventional methods depend on extensive linguistics knowledge, intricate feature extraction methods, and segmentation. In this paper, a novel method for handwriting identification using Connectionist Temporal Classification in conjunction with Convolutional Recurrent Neural Network is presented. The goal of this work is to categorize a single handwritten word so that handwritten material can be converted to digital format. Since each character’s occurrence is contingent upon the existence of a few preceding characters, the challenge involves both sequence and classification modeling. To create a model that can segment characters and reassemble each word, we look into different neural network topologies. Our model will be trained and tested using the IAM handwriting database. Additionally, the model may be trained globally and across many languages due to the symbolic and character-independent nature of the employed approaches.
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4401609142