Scholarlink

Wild animals are steadily destroying more crops close to forested regions, and vehicle-animal collisions are becoming a bigger risk to both people and wildlife. Techniques that are affordable for studying the behaviour of wild animals are required. This paper presents a YOLOv8n and ResNet18 model to predict and classify wild animals and give an alert to people near-by. The dataset is constructed from various sources includes Kaggle, and web scraping tool iCrawler.io. The proposed detection model yolov8n achieves a 71.8% mAP of 10 distinct classes and performance of the provided classification model, ResNet18 has significantly improved, with an average accuracy of 93.33% of 5 distinguished classes. This concept is a dependable way to preserve human life while delivering precise information based on animals.

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

Smart Wildlife Animal Tracking and Safety Alerts with Deep Learning Frameworks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • S. Pujitha,
  • G. Kalyani,
  • S. Rajasekhar,
  • B. Rohith Nayak

摘要

Wild animals are steadily destroying more crops close to forested regions, and vehicle-animal collisions are becoming a bigger risk to both people and wildlife. Techniques that are affordable for studying the behaviour of wild animals are required. This paper presents a YOLOv8n and ResNet18 model to predict and classify wild animals and give an alert to people near-by. The dataset is constructed from various sources includes Kaggle, and web scraping tool iCrawler.io. The proposed detection model yolov8n achieves a 71.8% mAP of 10 distinct classes and performance of the provided classification model, ResNet18 has significantly improved, with an average accuracy of 93.33% of 5 distinguished classes. This concept is a dependable way to preserve human life while delivering precise information based on animals.