About The Project
UTIClassification brings laboratory-grade AI assistance to microscopic urine images. The goal is to offer a transparent workflow for experimenting with computer-vision models that separate non-UTI cases from probable infections.
Guiding Pillars
Clinical Awareness
Highlight the prevalence of UTIs and empower clinicians, laboratorians, and patients with faster triage information.
Imaging + AI
Combine microscope imagery with deep-learning models to grade samples consistently and reduce manual review time.
Collaborative Tools
Provide an interface that researchers, students, and healthcare teams can iterate on together as datasets evolve.
Data & Models
The demo ships with two pretrained models: `UNU_ResNet101V2_Round5.keras` for Non-UTI detection and `UTI_VGG19_Round4.keras` for UTI classification. Each is exposed through the web interface so you can evaluate predictions without downloading tooling.
Want to contribute or extend the models?
Explore the repository, tweak the training notebooks, or adapt the API to your preferred deployment target.