Scholarlink

In this paper, the aim is to develop a strategy for the identification of animal species utilizing computer vision techniques. The proposed work is based on You only Look Once version 8 (YOLOv8), which emphasizes the algorithm's ability to identify objects in live video feeds or real-world scenarios which are efficiently used for wild animal detection in diverse habitats. The algorithm's speed and adaptability make it an ideal tool for the proposed work. Here, a comprehensive wildlife dataset has been considered for achieving accurate and rapid animal detection. The methodology described employs a deep learning framework to monitor wildlife in real time across images, videos, and live camera feeds. The dataset is compiled from a range of sources including documentaries, YouTube content, and preexisting datasets found on Internet sources, encompassing 7000 images with annotations for various object categories and highlights the use of artificial intelligence in tracking wildlife through different video sources. Multiple augmentation techniques were employed on the dataset images to enhance the model's precision. The proposed method attained an accuracy rate of 96.3%, showcasing its efficiency in identifying wild animals in real time at a rate of 20 frames per second (FPS).

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

Efficient Wild Animal Detection and Classification Using YOLOv8

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Raghavendra Reddy,
  • Duddukurireddy Pavankalyan,
  • S. U. Amulya,
  • Shaik Muzammil Irshad

摘要

In this paper, the aim is to develop a strategy for the identification of animal species utilizing computer vision techniques. The proposed work is based on You only Look Once version 8 (YOLOv8), which emphasizes the algorithm's ability to identify objects in live video feeds or real-world scenarios which are efficiently used for wild animal detection in diverse habitats. The algorithm's speed and adaptability make it an ideal tool for the proposed work. Here, a comprehensive wildlife dataset has been considered for achieving accurate and rapid animal detection. The methodology described employs a deep learning framework to monitor wildlife in real time across images, videos, and live camera feeds. The dataset is compiled from a range of sources including documentaries, YouTube content, and preexisting datasets found on Internet sources, encompassing 7000 images with annotations for various object categories and highlights the use of artificial intelligence in tracking wildlife through different video sources. Multiple augmentation techniques were employed on the dataset images to enhance the model's precision. The proposed method attained an accuracy rate of 96.3%, showcasing its efficiency in identifying wild animals in real time at a rate of 20 frames per second (FPS).