Scholarlink

<p>Hand sign recognition in the Mizo language using machine learning algorithms is a developing field. The primary challenge lies in lack of a standardized and comprehensive dataset for Mizo Sign Language (MSL), as sign language use can vary within Mizo households. However, research efforts are underway to address this gap, and the development of MSL datasets is a crucial first step. The aim of this study is to assist deaf and dumb people with Mizo language by hand sign recognition feature extraction and classification using machine learning algorithm. In this research the input is collected as hand sign of Mizo language dataset and processed for noise removal and segmentation. Then this segmented hand sign image features are extracted utilizing hidden Markov Conditional Context Generation attention neural random fields (HMCCGANRF). The extracted features are classified using transfer convolutional recurrent DenseNet (TCRDnet) model. From the classified results, separated specific Mizo sign is obtained and categorized. The experimental results have been carried out in terms of recognition accuracy, precision, recall, F-1 score and context sensitivity. Proposed technique attained Recognition accuracy of 97%, Precision of 95%, RECALL of 98%, Context sensitivity of 96%, F-1 SCORE of 94%.</p>

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

Hand sign recognition based on Mizo language using feature extraction and classification by machine learning algorithms

  • 期刊论文

  • 发表日期：

    2026年3月22日

查看全文
  • Barrister Ramsiej,
  • V. D Ambeth Kumar

摘要

Hand sign recognition in the Mizo language using machine learning algorithms is a developing field. The primary challenge lies in lack of a standardized and comprehensive dataset for Mizo Sign Language (MSL), as sign language use can vary within Mizo households. However, research efforts are underway to address this gap, and the development of MSL datasets is a crucial first step. The aim of this study is to assist deaf and dumb people with Mizo language by hand sign recognition feature extraction and classification using machine learning algorithm. In this research the input is collected as hand sign of Mizo language dataset and processed for noise removal and segmentation. Then this segmented hand sign image features are extracted utilizing hidden Markov Conditional Context Generation attention neural random fields (HMCCGANRF). The extracted features are classified using transfer convolutional recurrent DenseNet (TCRDnet) model. From the classified results, separated specific Mizo sign is obtained and categorized. The experimental results have been carried out in terms of recognition accuracy, precision, recall, F-1 score and context sensitivity. Proposed technique attained Recognition accuracy of 97%, Precision of 95%, RECALL of 98%, Context sensitivity of 96%, F-1 SCORE of 94%.