Scholarlink

<p>FR (face recognition) is the method to identify individuals via facial images. It is also important for surveillance and security applications and it is required in offices, social organizations and institutions. Conventional models have challenges such as pose variation, occlusion, and limited feature representation. These challenges disrupt their efficiency and scalability in real-world applications. In recent times, the advancement in deep learning (DL) has revolutionized FR by shifting from conventional models to local feature models, increasing robustness and accuracy. Hence, this work presents an automated FR model for optimization based DL model. The proposed FR has the following stages (a) image augmentation, (b) preprocessing, (c) optimal feature extraction, and (d) and optimal FR process. Initially, the input images are augmented and pre-processed. Then, the ODRCN (optimal deep residual capsule network) is used for FR process. Here, the hyperparameters of ORCN are optimized by the metaheuristic optimizer African vulture algorithm (AVA). The experimentation is performed on the LFW benchmark dataset and achieved better accuracy and FPR values of 99.4% and 0.9 respectively. Finally, it is proved the capacity of the proposed FR over the conventional FR models. These findings show the potential of the suggested ODRCN for real-time face recognition in resource-constrained environments.</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

An efficient face recognition model using an optimization based residual capsule network

  • 期刊论文

  • 发表日期：

    2025年7月21日

查看全文
  • B. Geetha Kumari,
  • Arpita Gupta

摘要

FR (face recognition) is the method to identify individuals via facial images. It is also important for surveillance and security applications and it is required in offices, social organizations and institutions. Conventional models have challenges such as pose variation, occlusion, and limited feature representation. These challenges disrupt their efficiency and scalability in real-world applications. In recent times, the advancement in deep learning (DL) has revolutionized FR by shifting from conventional models to local feature models, increasing robustness and accuracy. Hence, this work presents an automated FR model for optimization based DL model. The proposed FR has the following stages (a) image augmentation, (b) preprocessing, (c) optimal feature extraction, and (d) and optimal FR process. Initially, the input images are augmented and pre-processed. Then, the ODRCN (optimal deep residual capsule network) is used for FR process. Here, the hyperparameters of ORCN are optimized by the metaheuristic optimizer African vulture algorithm (AVA). The experimentation is performed on the LFW benchmark dataset and achieved better accuracy and FPR values of 99.4% and 0.9 respectively. Finally, it is proved the capacity of the proposed FR over the conventional FR models. These findings show the potential of the suggested ODRCN for real-time face recognition in resource-constrained environments.