LeafDiseaseDetection-DL
The original senior Bachelor's project: separate U-Net segmentation and MobileNetV2 classification models.
The original coursework project this portfolio's LeafLens project later redesigned. It trains two completely separate models: a U-Net for leaf segmentation and a MobileNetV2 classifier for disease identification. The two share nothing — not weights, not a pipeline, not an inference path.
That separation is exactly what motivated the LeafLens rework: training and serving two full backbones is wasteful, and knowing where the leaf is should help diagnose it. This project is kept here as the baseline that comparison is measured against, not as a live application — it's a training notebook and a written report, with no served inference API or UI.
What's here
Final_Project.ipynb— the full training and evaluation notebook for both models.Final_Project_Report.pdf— the written report accompanying the coursework submission.
View the source for the notebook and report.