Sloop: A pattern retrieval engine for individual animal identification

A computer vision for animal ecology

“…This approach was pioneered for fluke identification in marine mammals (Adams, Speakman, Zolman, & Schwacke, ; Beekmans, Whitehead, Huele, Steiner, & Steenbeek, ; Gilman, Hupman, Stockin, & Pawley, ) and has since been applied on a wide range of taxa, from zebras ( Equus grevyi ) (Crall, Stewart, Berger‐Wolf, Rubenstein, & Sundaresan, ), to elephants ( L. cyclotis ) (Ardovini, Cinque, & Sangineto, ), and box turtles ( Terrapene carolina ) (Cross, Lipps, Sapak, Tobin, & Root, ). These methods are effective in identifying animals with complex markings, such as giraffes ( Giraffa camelopardalis ) (Bolger, Morrison, Vance, Lee, & Farid, ), whale sharks ( Rhincodon typus ) (Arzoumanian et al., ) and catfish ( Rineloricaria aequalicuspis ) (Dala‐Corte, Moschetta, & Becker, ), and range from completely automated (Town, Marshall, & Sethasathien, ), to involving human feedback during matching (Duyck et al., ). Crall et al.…”

“…For animals with unique markings, this can be a low-cost alternative to expensive trapping and tagging programs. This approach was pioneered for fluke identification in marine mammals (Adams, Speakman, Zolman, & Schwacke, 2006;Beekmans, Whitehead, Huele, Steiner, & Steenbeek, 2005;Gilman, Hupman, Stockin, & Pawley, 2016) and has since been applied on a wide range of taxa, from zebras (Equus grevyi) (Crall, Stewart, Berger-Wolf, Rubenstein, & Sundaresan, 2013), to elephants (L. cyclotis) (Ardovini, Cinque, & F I G U R E 5 Application of computer vision to predicting individual and species identity: (a) Matching algorithms score the similarity of photographed zebras to a library of known images to track individuals over time (Crall et al 2013 (Town, Marshall, & Sethasathien, 2013), to involving human feedback during matching (Duyck et al, 2015). Crall et al (2013) reported accuracy rates ranging from 95% for Grey's zebras (E. grevyi) to 100% for jaguars (Panthera onca) using the HotSpotter algorithm, which can be accessed through the Wildbook web platform.…”

Survival of a grey reef shark Carcharhinus amblyrhynchos without a dorsal fin

“…In turn, these results do not imply that shark finning will have negligible consequences. Hopefully the rising popularity of citizen‐science programmes and improving algorithms for recognising sharks from photographs (Duyck et al, ) will allow a step change in the measurement of shark survival and eventually allow the demographic consequences of injuries to be estimated for multiple species.…”

Crowdsourcing Fundamentals: Definition and Typology

“…Other applications are establishing the fingerprint of past sea-level changes (Rovere et al 2012), validate data to generate overall landcover maps (See et al 2014), urban surveillance (Monahan and Mokos 2013), or animal identification for ecological monitoring and conservation (Duyck et al 2014). …”