Scholarlink

For a postal document to be delivered to the intended location, it is critical to correctly identify the destination city name. When writing addresses in India, individuals sometimes mix up the scripts. The destination city’s name is frequently written differently from the other parts of the postal document. India’s multilingual and multi-script culture makes this widespread there. This paper presents a Convolutional Neural Network (CNN) based method for handwritten multilingual multi-script Indian city name recognition. Experiments were conducted using English, Bangla, and Devanagari scripts in addition to a single script scenario. The proposed technique produced an accuracy of 91.72% on 106 city names in a mixed script scenario and on the behalf of single script accuracy presented as 96.27% Bangla, 93.30% in English, and 98.20% in Devanagari.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Recognition of Handwritten Multilingual City Names Using CNN

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Harikesh Pandey,
  • Nidhi Gupta,
  • Arun Prakash Agrawal

摘要

For a postal document to be delivered to the intended location, it is critical to correctly identify the destination city name. When writing addresses in India, individuals sometimes mix up the scripts. The destination city’s name is frequently written differently from the other parts of the postal document. India’s multilingual and multi-script culture makes this widespread there. This paper presents a Convolutional Neural Network (CNN) based method for handwritten multilingual multi-script Indian city name recognition. Experiments were conducted using English, Bangla, and Devanagari scripts in addition to a single script scenario. The proposed technique produced an accuracy of 91.72% on 106 city names in a mixed script scenario and on the behalf of single script accuracy presented as 96.27% Bangla, 93.30% in English, and 98.20% in Devanagari.