Combining imperfect automated annotations of underwater images with human annotations to obtain precise and unbiased population estimates

详细信息    查看全文

• 作者：Jui-Han Chang^a ; Deborah R. Hart^a ; ^{deborah.hart@noaa.gov} ; Burton V. Shank^a ; Scott M. Gallager^b ; Peter Honig^b ; Amber D. York^b
• 关键词：Underwater imagery ; Computer vision ; Population estimation ; Scallop ; Geographically weighted regression
• 刊名：Methods in Oceanography
• 出版年：2016
• 出版时间：December 2016
• 年：2016
• 卷：17
• 期：Complete
• 页码：169-186
• 全文大小：1887 K
• 卷排序：17

文摘

Methods for estimating population size from imperfect image data are analyzed. Stratification with a local ratio or regression estimator is generally most effective. The false negative rate is key in determining the best method.