Journal of Applied Optics, Volume. 44, Issue 2, 323(2023)
Weather recognition method based on convolutional neural network and feature fusion
Figures & Tables(13)
Fig. 4. Image shape feature diagram under different weather conditions
Fig. 7. Relationship between training iteration times and accuracy
Table 1. Experimental hardware and software environment
View tableView in ArticleTable 1. Experimental hardware and software environment
硬件/软件 型号 处理器 AMD Ryzen 7 5800H 内存 2×16 GB DDR4 显卡 NVIDIA Geforce RTX 3060 硬盘 1T SSD PCle NVMe TLC 操作系统 Windows 框架 TensorFlow CUDA CUDA 11.1 Anaconda Anaconda3-5.2.0(python3.6)
Table 2. Experimental results of test set
View tableView in ArticleTable 2. Experimental results of test set
天气(样本数) 识别结果 正确率/% 晴天 阴天 雨天 雪天 晴天(800) 745 34 0 21 93.1 阴天(800) 26 753 12 9 94.1 雨天(800) 18 12 770 0 96.3 雪天(800) 34 18 7 741 92.6
Table 3. Experimental results of validation set
View tableView in ArticleTable 3. Experimental results of validation set
天气(样本数) 识别结果 正确率/% 晴天 阴天 雨天 雪天 晴天(20) 19 1 0 0 95.0 阴天(20) 0 20 0 0 100.0 雨天(20) 0 1 19 0 95.0 雪天(20) 0 0 0 20 100.0
Tools
Get Citation
Copy Citation Text
Xiaoming HOU, Yafeng QIU. Weather recognition method based on convolutional neural network and feature fusion[J]. Journal of Applied Optics, 2023, 44(2): 323
Paper Information
Category: Research Articles
Received: May. 11, 2022
Accepted: --
Published Online: Apr. 14, 2023
The Author Email: QIU Yafeng (njlghcn@sina.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20