Journals
  Publication Years
  Keywords
Search within results Open Search
Please wait a minute...
For Selected: Toggle Thumbnails
Instance Segmentation of Buildings from High-Resolution Remote Sensing Images with Multitask Learning
HUI Jian, QIN Qiming, XU Wei, SUI Juan
Acta Scientiarum Naturalium Universitatis Pekinensis    2019, 55 (6): 1067-1077.   DOI: 10.13209/j.0479-8023.2019.106
Abstract2256)   HTML    PDF(pc) (22062KB)(410)       Save
At present, building extraction from high-resolution remote sensing images using deep neural network is viewed as a binary classification problem, which divides the pixels into two categories, building and nonbuilding, but it cannot distinguish individual buildings. To solve this problem, the U-Net modified with Xception module and multitask learning are combined to apply to the instance segmentation of buildings, which both acquires the binary classification and distinguishes the individual buildings. Inria aerial imagery is used as the research dataset to validate the algorithm. The results show that the binary classification performance of U-Net modified with Xception outperforms U-Net by about 1.4%. The multitask driven deep neural network not only accomplishes the instance segmentation of buildings, but also improves the accuracy by about 0.5%.
Related Articles | Metrics | Comments0
Ground-Based Measurement and Variation Analysis of Carbonaceous Aerosols in Wuqing
XU Wei, FU Tzung-May, CHEN Jinxuan, TIAN Heng
Acta Scientiarum Naturalium Universitatis Pekinensis    2016, 52 (3): 409-419.   DOI: 10.13209/j.0479-8023.2015.144
Abstract1146)   HTML    PDF(pc) (875KB)(955)       Save

To understand the concentrations and sources of carbonaceous aerosols in Northern China, real-time, semi-online, hourly measurements of PM2.5 compositions were conducted at an urban site in Wuqing, Tianjin from December 31, 2011 to January 11, 2012. The mean concentrations of EC and OC in Wuqing were 6.0±4.8 and 21.5±19.2 μg C/m3 respectively, which constituted 8% and 30% of the total measured PM2.5 constituent mass. The mean concentration of WSOC was 14.3±11.8 μg C/m3, which constituted 67% of the mean OC concentration. During the observation period, the large variability of pollutant concentrations were mainly driven by synopticscale meteorological events. As a result, the diurnal patterns of EC, OC, and WSOC were relatively indistinct. The observed mass ratios of OC/EC was relatively stable throughout the observation period and averaged 3.9. Based on correlation analysis with other tracer constituents, it is found that the wintertime carbonaceous aerosols in Wuqing came mainly from biomass burning emissions and experienced significant aging. Roughly half of the OC were from biomass burning; the other half were from secondary formation processes.

Related Articles | Metrics | Comments0