Scholarlink

<p>The paper is interested in detecting the estrus status of dairy cows from videos of a cattle. A complete fully image-driven processing chain is developed for estrus classification. First, the cow detector is performed by the lightweight deep neural network YOLOv8n Pose which also provides the keypoints of the cows in each frame. The tracking of the cows is carried out by the BotSort tracker. Spatio-temporal features are then computed around the keypoints of each cow. These descriptors are the input of a binary classifier that detects the estrus status and, a benchmark of several classifiers is conducted. Experimental results indicate that our method outperforms the state of art methods while using fewer parameters, validating the benefits of a lightweight and adaptive tracking solution.</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 improved cow estrus detection system based on computer vision

  • 期刊论文

  • 发表日期：

    2026年5月29日

查看全文
  • Roua Mkadmi,
  • Rabaa Youssef-Douss,
  • Amel Benazza-Benyahia

摘要

The paper is interested in detecting the estrus status of dairy cows from videos of a cattle. A complete fully image-driven processing chain is developed for estrus classification. First, the cow detector is performed by the lightweight deep neural network YOLOv8n Pose which also provides the keypoints of the cows in each frame. The tracking of the cows is carried out by the BotSort tracker. Spatio-temporal features are then computed around the keypoints of each cow. These descriptors are the input of a binary classifier that detects the estrus status and, a benchmark of several classifiers is conducted. Experimental results indicate that our method outperforms the state of art methods while using fewer parameters, validating the benefits of a lightweight and adaptive tracking solution.