| 摘 要: 针对鸟类的识别对人类和计算机视觉要求较高，视频检测和人工识别等传统的识别方法均存在一定的困难，提出一种基于深度学习的鸟类识别系统，借助百度EasyDL深度学习平台，以南方野生鸟类为实验对象进行数据标注、模型建立、模型训练，经验证集测试验证，识别精确率为94.2%。文章提出的模型可应用于已知鸟类的实时监测识别，并辅助实现对未知鸟类的监测与发现，为生物多样性保护提供技术支撑。 |
| 关键词: 深度学习；百度EasyDL；野生鸟类；物体检测；模型 |
|
中图分类号: TP391
文献标识码: A
|
| 基金项目: deep learning; Baidu EasyDL; wild birds; object detection; model |
|
| Design of Southern Wild Bird Recognition System based on Deep Learning |
|
LAI Xiaoping
|
(College of Information Technology, Guangdong Communication Polytechnic, Guangzhou 510650, China )
laixiaoping1008@163.com
|
| Abstract: Recognition of birds requires a high level of human and computer vision, and traditional recognition methods such as video detection and manual recognition have certain difficulties. In view of these problems, this paper proposes a bird recognition system based on deep learning. With the help of Baidu EasyDL depth learning platform, data annotations, model building and model training are carried out by taking southern wild birds as experimental objects. Verification set test results show that the recognition accuracy rate is 94.2%. The proposed model can be applied to the real-time monitoring and recognition of the existing birds, and assist in the monitoring and discovery of unknown birds, providing technical support for biodiversity conservation. |
| Keywords: deep learning; Baidu EasyDL; wild birds; object detection; model |
