We propose a scalable approach to learn video-based question answering (QA): answer a "free-form natural language question" about a video content. Our approach automatically harvests a large number of videos and descriptions freely available online. Then, a large number of candidate QA pairs are automatically generated from descriptions rather than manually annotated. Next, we use these candidate QA pairs to train a number of video-based QA methods extended fromMN (Sukhbaatar et al....
Topics: Computer Vision and Pattern Recognition, Artificial Intelligence, Multimedia, Computing Research...
Source: http://arxiv.org/abs/1611.04021

We estimate the global Hg wildfire emissions for the 2000s and the potential impacts from the 2000–2050 changes in climate, land use and land cover and Hg anthropogenic emissions by combining statistical analysis with global data on vegetation type and coverage as well as fire activities. Global Hg wildfire emissions are estimated to be 612 Mg year −1 . Africa is the dominant source region (43.8% of global emissions), followed by Eurasia (31%) and South America (16.6%). We find significant...
Topics: Biomass burning, Climate change, Land cover, Land use, Modeling

Particulate matter (PM) in the atmosphere has adverse effects on human health, ecosystems, and visibility. It also plays an important role in meteorology and climate change. A good understanding of its sources is essential for effective emission controls to reduce PM and to protect public health. In this study, a total of 239 PM source apportionment studies in China published during 1987–2017 were reviewed. The documents studied include peer-reviewed papers in international and Chinese...
Topics: China, PM 2.5, receptor models, regional variations