Applying computer vision to natural history collections for ecological, taxonomic and conservation research
Exploring the vast amount of computer vision data extracted from natural history collections.
Recent mass digitization efforts of Natural History Collections (NHCs), pioneered by institutions like the Natural History Museum, have given scientists across the globe access to untold numbers of specimens and associated meta-data (e.g. date and location of collection) for research.
Using these digitized collections, the Natural History Museum and colleagues are developing new computer vision techniques to automatically extract biological and historical information from NHCs across the tree of life and from a wide array of habitats. Crucially however, these massive datasets must now be focused towards unlocking their potential for answering ecological, taxonomic and conservation questions.
This studentship will explore the vast amount of computer vision data extracted from NHCs and apply them to the following questions:
- Can computer vision be used to fill taxonomic gaps and facilitate new species descriptions?
- Can the morphological features (e.g. body size, shape, colour) extracted from computer vision be used to examine biotic response to climate change (e.g. temperature-size responses)?
- Can computer vision be used as a means for rapid biodiversity assessments of field collected invertebrates from a wide range of habitats (e.g. forests to the deep sea)?
This PhD will be among the first to apply computer vision methods to taxonomic, ecological, and conservation questions.
The PhD student will use biological data extracted from specimens using computer vision techniques and apply them to targeted research questions that could not be adequately answered using traditional methods (i.e. manual measurements of specimens).
The biological data will include: morphological features (e.g. body size/mass, colour, shape, sex, and other phenotypic patterns) and specimen level meta-data (date and location of collections). Specimens will be drawn from those housed at the Natural History Museum, the Discovery Collections (NOC), and from those collected in the field by the PhD student.
The student will use these data (along with targeted DNA barcoding) to identify specimens from regions with poor taxonomic coverage (e.g. tropical regions/deep sea) to uncover previously unrecognized between and within species diversity – with the potential for describing new species and uncovering phenotypic variation within species.
In addition, the student will pair historic temperature and land use records with automated measurements of morphological features of NHCs to test hypotheses (e.g. the temperature-size rule) surrounding biotic response to climate change and other human impacts (habitat degradation) over long time scales (100 years+).
Finally, the student will extract morphological data from bulk samples of terrestrial and marine invertebrates to develop rapid methods for biodiversity assessment.
The INSPIRE DTP programme provides comprehensive personal and professional development training alongside extensive opportunities for students to expand their multi-disciplinary outlook through interactions with a wide network of academic, research and industrial/policy partners.
The student will be registered and hosted at the University of Southampton (NOC) but there will be a significant amount of research to be conducted at the Natural History Museum.
Specific training will include:
- digitization of natural history collections
- collections and data management
- DNA barcoding
- machine learning
- phylogenetic and taxonomic analysis
- public speaking and written communication.
Eligibilily and how to apply
Please check the INSPIRE website for details on how to apply.
Application deadline: 4 January 2021.
Wilson RJ, Brooks SJ, Van der Walt SJ, Feng D, Price BW, Fenberg PB (submitted) Temperature-size responses in British Lepidoptera using computer vision applied to natural history collections (manuscript available upon request).
Hansen OLP, Svenning JC, Olsen K, Dupont S, Garner BH, Iosifidis A, Price BW and Høye TT (2020) Species-Level Image Classification with Convolutional Neural Network Enables Insect Identification from Habitus Images. Ecology and Evolution 10(2):737-47.
Wonglersak R, Fenberg PB, Langdon PG, Brooks SJ, Price BW (2020) Temperature-body size responses in insects: a case study of British Odonata. Ecological Entomology 45(4): 795-805.
This a joint PhD training partnership between the Natural History Museum and INSPIRE a NERC Doctoral Training Partnership (DTP) creating an innovative multi-disciplinary experience for the effective training of future leaders in environmental science, engineering, technology development, business, and policy.