HotSpotter: Using a computer-driven photo-id application to identify sea turtles

Dunbar, Stephen G.; Anger, Edward C.; Parham, Jason; Kingen, Colin; Wright, Marsha K.; Hayes, Christian; Safi, Shahnaj Binte; Holmberg, Jason; Salinas, Lidia A.; Baumbach, Dustin S.

doi:10.1016/j.jembe.2020.151490

Cited by 28 publications

(25 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This study also took advantage of social media mining which together with dedicated citizen science platforms can help towards the collection of imaging data in large spatial and temporal scales and also retrospectively (Papafitsoros et al, 2021;Papafitsoros et al, 2022). Recent and continuous developments in automatic animal re-identification algorithms (Dunbar et al, 2021;Stewart et al, 2021) in combination with automatic mining for animal imagery in social media, open-data citizen science platforms (Berger-Wolf et al, 2017;Giovos et al, 2019), as well as advances in machine learning already facilitate these type of studies (Di Minin et al, 2018;Dujon and Schofield, 2019) and are expected to do so even more in the future.…”

Section: Greecementioning

confidence: 99%

Social media mining and photo-identification detects the shift of long-term seasonal foraging habitat for a juvenile loggerhead sea turtle

Papafitsoros

2022

Preprint

View full text Add to dashboard Cite

Animal imagery uploaded on social media has been identified as an important tool in wildlife research and conservation and has been used in a variety of recent studies. Here a case study is presented, where a detailed analysis of social media content revealed a shift of a long-term foraging habitat for a juvenile loggerhead sea turtle (Caretta caretta) in Greece. In particular, this individual was a long term resident of Zakynthos island, Greece, from 2016 until 2020 (5 consecutive summer seasons), regularly foraging on a nearshore reef, with no observations of it being made during the 2021 season. However regular social media image mining combined with photo-identification, detected this individual foraging in the Gulf of Corinth in August 2021, more that 200km away from his previous foraging habitat. This case study (i) shows the possibility for juvenile loggerheads to shift their foraging site even after long term use, with implications in capture-mark-recapture studies and (ii) once more highlights the useful- ness of social media mining and citizen science in diverse aspects of sea turtle studies.

show abstract

Section: Greecementioning

confidence: 99%

Social media mining and photo-identification detects the shift of long-term seasonal foraging habitat for a juvenile loggerhead sea turtle

Papafitsoros

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Adobe Lightroom R was used in some cases to enhance photographs making it easier for turtle identification. Photographers were instructed to focus on capturing the left and right head profiles of each individual in order to target the post-orbital scales, which are the primary focal point of most sea turtle individual recognition software packages (e.g., Dunbar et al, 2014Dunbar et al, , 2021. Photographs that incorporated portions of the head, flippers or carapace were also useful to distinguish individuals.…”

Section: Photography and Digital Image Recognitionmentioning

confidence: 99%

“…With the growth of wildlife photography, due in part to advanced camera and mobile technology, the role of local communities in scientific data collection via photography has expanded (e.g., Azzurro et al, 2013;Baumbach et al, 2019). Participating citizens can now download various applications onto their mobile devices allowing them to submit photographs along with time and GPS location data (Michonneau and Paulay, 2015;Baumbach and Dunbar, 2017;Papafitsoros et al, 2020;Dunbar et al, 2021). Citizens are able to contribute to scientific research as they go about their daily life, while exploring local ecosystems, or when traveling to remote locations (Graham et al, 2011;Read and Jean, 2021).…”

Section: Introductionmentioning

confidence: 99%

“…In addition, there are many sea turtle foraging sites (e.g., Great Barrier Reef, Cayman Islands) that overlap with areas that have high amounts of recreational activities such as snorkeling, diving, kayaking and paddle boarding (Baumbach et al, 2019), and are thus may be inappropriate for hosting invasive turtle capture efforts. In these areas, citizen science has provided unique opportunities to gather large amounts of data in a minimally intrusive way while also educating the public about local species and their ecology (Bell et al, 2008;Hof et al, 2017;Nordstrom et al, 2019;Dunbar et al, 2021). However, maximizing the value of citizen science requires some level of engagement with professional scientists or researchers to help orient and inform best practices for data collection.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Citizen-Sourced Sightings and Underwater Photography Reveal Novel Insights About Green Sea Turtle Distribution and Ecology in Southern California

et al. 2021

View full text Add to dashboard Cite

East Pacific (EP) green turtles (Chelonia mydas) have undergone substantial population recovery over the last two decades owing to holistic protection at nesting beaches and foraging areas. At the northern end of their range in southern California United States, green turtles have been seen in more areas and in greater numbers since 2014 than before as a result. A resident population of green turtles has established near La Jolla Shores (LJS), a protected site with daily marine tourism (e.g., kayakers, snorkelers, divers). To study this local aggregation, innovative and non-invasive methods were required because the traditional capture-recapture methods were infeasible due to public relations sensitivities. Green turtle habituation to humans at this site has created a unique opportunity for citizen-based science using underwater photography to document turtles and their surroundings. We obtained 309 usable photographs of local green turtles from members of the dive/snorkel community in LJS. Photos were taken from April 2016 to June 2019. Images were processed in Hotspotter—a patterned species instance recognition software—to identify seven individuals, five of which were consistently photographed throughout that period. These images helped infer minimum residency duration (MRD), seasonal differences in algal coverage on the carapace, habitat association, behavioral patterns, and diet. Mean MRD was 424 days (SE = 131 days, calculated from entire population, n = 7), during which turtles were active in 82.8% of the photographs; the remainder of the photographs depicted foraging (14.9%) or resting behavior (2.3%). Green turtles were seen foraging in water temperatures as low as 15.8°C, the lowest recorded temperature for foraging green turtles documented in literature. Additional opportunistic observational platforms were used to look at trends of increasing green turtle abundance in southern California since 2015 that supported the arrival of a new aggregation of green turtles in LJS. Our use of citizen-sourced photographs confirms the presence of a resident aggregation of green turtles in LJS. Existence of green turtles and other protected species in highly populated areas provide excellent opportunities to educate beachgoers and seafarers about conservation of these species. This study also highlights the value of citizen-based science in areas where traditional research techniques are ill-suited.

show abstract

“…Hotspotter was developed from the SIFT and Local Naive Bayes Nearest Neighbor (LNBNN) algorithms. It has also shown high reliability in fishes (Crall et al, 2013), mollusks (Barord et al, 2014), insects (Quinby, Creighton & Flaherty, 2021), reptiles (Dunbar et al, 2021 ;Tabuki et al, 2021) and mammals (Crall et al, 2013 ;Lea et al, 2018 ;Nipko, Holcombe & Kelly, 2020). Not all individually identifiable patterns, however, are stable over time.…”

Section: Introductionmentioning

confidence: 99%

Rapid colour changes in a tiny threatened gecko do not impede computer-assisted individual recognition

Monnet

Dokhelar

Renet

2022

Preprint

View full text Add to dashboard Cite

Photo-identification is a non-invasive method used for recognising wild animals with distinctive and stable patterns over time. This method is now widely used for capture-recapture wildlife monitoring. However, in species exhibiting rapid colouration changes, the evolving body patterns can lead to errors in individual recognition. In this study, we assessed the effect of dorsal physiological colour change of the tiny threatened European leaf-toed gecko (Euleptes europaea) on the performance of Wild-ID and Hotspotter, the two most commonly used individual recognition software for wildlife monitoring. We exposed 30 European leaf-toed geckos to several semi-controlled parameters (substrate type, temperature and light from natural diurnal/nocturnal cycles) in order to characterise the extent of variation in dorsal colouration, by standardised reflectance measurements. The colour of the substrate had a significant effect on individual reflectance changes. Body temperature also seemed to significantly affect the reflectance but the experimental conditions did not allow us to clearly distinguish the effect of temperature and light. For each of the 30 geckos, four photographic databases (n = 4*280) were then analysed by both software packages, under two extreme reflectance conditions. Despite the large changes in individual reflectance, Wild-ID and Hotspotter proved to be extremely reliable with a 100% recognition rate. The analysis of similarity scores suggests that Hotspotter is less sensitive to chromatic variation than Wild-ID. We provide here the first evidence that physiological colour change is not a barrier to computer-assisted individual recognition. This study advocates the use of Hotspotter for monitoring populations of European leaf-toed geckos and other saurians that generate significant colouration change over a short time.

show abstract

HotSpotter: Using a computer-driven photo-id application to identify sea turtles

Cited by 28 publications

References 35 publications

Social media mining and photo-identification detects the shift of long-term seasonal foraging habitat for a juvenile loggerhead sea turtle

Social media mining and photo-identification detects the shift of long-term seasonal foraging habitat for a juvenile loggerhead sea turtle

Citizen-Sourced Sightings and Underwater Photography Reveal Novel Insights About Green Sea Turtle Distribution and Ecology in Southern California

Rapid colour changes in a tiny threatened gecko do not impede computer-assisted individual recognition

Contact Info

Product

Resources

About