Scholarlink

<p>Precision livestock farming (PLF) leverages advanced technologies such as big data, the Internet of Things (IoT), machine learning, and deep learning to enhance livestock productivity and facilitate decision-making. This paper focuses on goat face detection and identification, crucial for intelligent goat management. Current methods face challenges in addressing the complex geometric transformations and variability in coat color of goat faces. To address these issues, we propose the Context-Assisted Astrous Deformable Network (CAADNet), which integrates a context-assisted module and astrous deformable convolution to model facial geometric transformations and exploit contextual information. CAADNet can be seamlessly integrated into existing networks, enhancing their representation capacity for goat face detection and identification. Extensive experiments on various detectors and backbones demonstrate the effectiveness of our approach, achieving state-of-the-art performance with 91.1% recall and 86.0% mAP on the goat face detection task and 98.0% identification accuracy on the goat face identification task. Our work reduces labor costs, provides a new method for accurate goat face detection and identification, and supports related research in precision livestock farming. The code is available at <a href="https://github.com/tiana-tang/Goat-Face-Detection-and-Recognition.git">https://github.com/tiana-tang/Goat-Face-Detection-and-Recognition.git</a>.</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

Context-assisted astrous deformable convolution for robust goat face detection and identification

  • 期刊论文

  • 发表日期：

    2025年12月17日

查看全文
  • Gaoge Han,
  • Lianyue Zhang,
  • Zihan Bai,
  • Xue Zhang,
  • Ruizi Han,
  • Chao Tang,
  • Jinglei Tang

摘要

Precision livestock farming (PLF) leverages advanced technologies such as big data, the Internet of Things (IoT), machine learning, and deep learning to enhance livestock productivity and facilitate decision-making. This paper focuses on goat face detection and identification, crucial for intelligent goat management. Current methods face challenges in addressing the complex geometric transformations and variability in coat color of goat faces. To address these issues, we propose the Context-Assisted Astrous Deformable Network (CAADNet), which integrates a context-assisted module and astrous deformable convolution to model facial geometric transformations and exploit contextual information. CAADNet can be seamlessly integrated into existing networks, enhancing their representation capacity for goat face detection and identification. Extensive experiments on various detectors and backbones demonstrate the effectiveness of our approach, achieving state-of-the-art performance with 91.1% recall and 86.0% mAP on the goat face detection task and 98.0% identification accuracy on the goat face identification task. Our work reduces labor costs, provides a new method for accurate goat face detection and identification, and supports related research in precision livestock farming. The code is available at https://github.com/tiana-tang/Goat-Face-Detection-and-Recognition.git.