On 13 – 16 November 2016, Pulse Lab Jakarta organized a Research Dive on Image Mining for Disaster Management, hosting 16 researchers from 14 universities across Indonesia. The participants worked in teams to develop analytical tools and generate research insights into four areas. During the two research days, researchers explored and analysed the data, guided by image processing and GIS senior researchers as advisors, and representatives from UN OCHA and the National Disaster Management Agency (BNPB) as domain experts on disaster management. This technical report presents the extended abstracts submitted by the four groups involved in the Research Dive on: i) generating automatic descriptions from images related to the haze crisis using deep learning, ii) inferring the level of visibility from hazy images by applying single-image and learning-based approaches, iii) measuring the impacts of volcanic eruption by using spatial data analysis and image processing approaches for satellite imagery data, iv) exploring the combination of inundation modeling and image mining to portray the relationship between vulnerability and flood inundation.
UN Global Pulse Finland