RSCM: Region Selection and Concurrency Model for Multi-class Weather Recognition

IEEE Transactions on Image Processing 2017

Di Lin, Cewu Lu, Member, IEEE Hui Huang, Member, IEEE Jiaya Jia, Senior Member, IEEE

Fig. 1. Regions of meadow in sunny and cloudy days in (a) and (b)respectively.


Abstract 

Towards weather condition recognition, we emphasize the importance of regional cues in this paper and address a few important problems regarding appropriate representation, its differentiation among regions, and weather-condition feature construction. Our major contribution is, first, to construct a multi-class benchmark dataset containing 65,000 images from 6 common categories for sunny, cloudy, rainy, snowy, haze and thunder weather. This dataset also benefits weather classification and attribute recognition. Second, we propose a deep learning framework named region selection and concurrency model (RSCM) to help discover regional properties and concurrency.We evaluate RSCM on our multi-class benchmark data and another public dataset for weather recognition.


Index Terms: Deep Learning, Multi-class Weather Recognition, Image Classification, Attribute Recognition.


Fig. 2. Overview of our multi-class weather dataset. It includes 6 weather categories, i.e., sunny, cloudy, rainy, snowy, haze and thunder.



Acknowledgments

This work is supported by a grant from the Research Grants Council of the Hong Kong SAR (project No. 2150760) and the National Natural Science Foundation of China (project No. 61472245). It was also supported in parts by the NSFC (61522213), 973 Program (2015CB352501), Guangdong Science and Technology Program (2015A030312015, 2016A050503036) and Shenzhen Innovation Program (JCYJ20151015151249564). Cewu Lu is the corresponding author of this paper.


Bibtex

@article{RSCM2017,
     title = {RSCM: Region Selection and Concurrency Model for Multi-class Weather Recognition},
     author = {Di Lin and Cewu Lu and Hui Huang and Jiaya Jia},
     journal = {IEEE Transactions on Image Processing},
     volume = {26},
     number = {9},
     year = {2017},
     pages = {4154--4167},   
}


Copyright © 2016-2018 Visual Computing Research Center