All projects
Machine Learning
WildlifeReID-10k Comparison
Group project comparing two models for recognising individual animals from photos.
- Role
- Group project member
PythonPyTorchFlaskReact
A Big Data Analytics group project (four students) on animal re-identification: deciding whether a photo shows a specific, previously seen individual animal, rather than just its species.
We compare MiewID, a wildlife-specific embedding model, with a general-purpose ConvNeXt Base model on the WildlifeReID-10k benchmark (over 140,000 images of more than 10,000 individuals across roughly 33 species), using the same experimental protocol for both. The aim is a system that returns ranked identity matches or an unknown-individual result, served through a REST API to a web app built with Flask and React.