Wei-Ta Chu, Xiang-You Zheng, and Ding-Shiuan Ding
Multimedia Computing Laboratory
Dept. of Computer Science and Information Engineering
National Chung Cheng University
1. Introduction
We estimate weather information from single images, as an important clue to unveil real-world characteristics available in the cyberspace, and as a complementary feature to facilitate computer vision applications. Based on an image collection with geotags, we crawl the associated weather and elevation properties from the web. With this large-scale and rich image dataset, various correlations between weather properties and metadata are observed, and are used to construct computational models based on random forests to estimate weather information for any given image. We describe interesting statistics linking weather properties with human behaviors, and show that image's weather information can potentially benefit computer vision tasks such as landmark classification. Overall, this work proposes a large-scale image dataset with rich weather properties, and provides comprehensive studies on using cameras as weather sensors.
2. Image2Weather Dataset
The following table shows statistics of the Image2Weather dataset, including numbers of photos in five different weather types. This dataset consists of 183,798 images in total. Overall, geographically the collected images span from 9.25 degrees west longitude to 30.4 degrees east longitude, and from 35 degrees north latitude to 62.21 degrees north latitude (covering most part of the Europe). The range of temperature is from -25 degree Celsius to 45 degree Celsius, and the range of humidity is from 4% to 100%.
Number of sunny images | 70,501 |
Number of cloudy images | 45,662 |
Number of snowy images | 1,252 |
Number of rainy images | 1,369 |
Number of foggy images | 357 |
Number of other images | 64,657 |
Total | 183,798 |
* Note that the distribution of the numbers of images in different weather conditions is different from the one we wrote in our paper (mentioned below), because we further manually correct weather types and label images with unclear weather conditions into "others".
Dataset download:
3. Citation
Please cite our work if you utilize this dataset.
Wei-Ta Chu, Xiang-You Zheng, and Ding-Shiuan Ding, "Camera as Weather Sensor: Estimating Weather Information from Single Images," Journal of Visual Communications and Image Representation, vol. 46, pp. 233-249, 2017.
DOI information: 10.1016/j.jvcir.2017.04.002
http://dx.doi.org/10.1016/j.jvcir.2017.04.002
Last Updated: April 25, 2017