Using AI Image Detection to Segment Fossil Material
Research Group Activity
The Centre of Imaging, Metrology, and Additive Technologies (CiMAT) specialises in employing cutting edge imaging techniques in the non-destructive characterization of materials across many discrete disciplines, including cultural heritage, forensics, automation, and manufacturing.
Mark A Williams leads the group, assisting academic and industrial partners on a huge diversity of project applications, with a particular focus on applying X-ray Computed Tomography (XCT) to solve real-world issues.
Project Description
The project will provide the student with the opportunity to study AI methods and refine their skills within image processing. XCT has become a core technology for describing fossil material and discovering new material that can yield key insights into life in the distant past. However, fossil material can be notoriously difficult to scan, often with low image contrast between the fossil and its surrounding matrix.
The result is the extremely time-consuming practice of manual segmentation, which can take weeks to months to process a single dataset. This is less than ideal. More advanced image processing methods could be employed to streamline this process into something more manageable.
The project will look at programming and scripting a solution to being able to image some problematic fossil material from the Cosely and Herefordshire Lagerstaette (a site of incredible preservation), defining the feasibility of employing such technology.
Required Skills
The project will require the student to already have a firm grasp of image processing in MATLAB or Python. The student will be expected to carry out research into AI methods and will be given space to study these and come up with solutions to the issue. An interest in non-destructive testing using CT and palaeontology are not necessary, but welcome!