| Summary: | 碩士 === 國立中正大學 === 資訊工程研究所 === 102 === We propose to estimate weather information from a single image, as an important clue to unveil real-world characteristics available in the cyberspace, and as a complementary attribute to facilitate computer vision applications. Based on a large-scale image collection with GPS information, we crawl weather and elevation properties associated with each image from a web-based weather report platform and a web-based map service, respectively. With this cross-platform heterogeneous 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.
|
|---|
