A DEEP LEARNING BASED METHOD FOR TYPHOON RECOGNITION AND TYPHOON CENTER LOCATION

As typhoons cause serious damages to human society, it is essential to recognize and locate typhoons. However, traditional methods depend heavily on handcrafted features and lack generalization ability. With the development of artificial intelligence, deep learning has greatly improved image recognition technology. To this end, we present a novel deep learning based method for typhoon recognition and typhoon center location. The proposed method utilizes a classification network (ClaNet) to determine if there is a typhoon in the given infrared (IR) satellite image, and a location network (LocNet) to locate the typhoon center if the classification result is positive. To train the proposed networks, we build a dataset of IR satellite images using data from Himawari-8 (H8). Moreover, we utilize class activation maps (CAM) to better understand the characteristics of this learning based method. Experimental results demonstrate that the proposed method can recognize typhoons effectively and locate typhoon centers accurately.