Tomato Disease Detection
Photograph a leaf, find out what's wrong with the plant — no agronomist required.
- Client
- Personal project
- Year
- 2024
- Status
- Source available
- Service
- Data quality & pipeline QA
- Model
- CNN (PyTorch)
- Input
- Leaf photograph
- Delivery
- Django web app
The problem
Identifying crop disease from a leaf normally needs expertise that isn't available in the field, so problems get caught late — after the damage spreads.
What I did
- Trained a convolutional neural network in PyTorch to classify leaf images by disease
- Handled the image preprocessing and augmentation pipeline with NumPy
- Wrapped the trained model in a Django app so a photo upload returns a prediction
The outcome
A working image-classification web app that turns a phone photo into a disease prediction — my first end-to-end machine-learning deployment.
Hire me for this
Data quality & pipeline QA
Reports are only as trustworthy as the data underneath them. I check that your numbers are actually right — across systems, after every update — and set up tests that catch it when they stop being right.
More work
- Sprachen Zentrum LMSA language school moved its whole course library online — without its videos leaking.2025
- Tathyanka TrackerNobody has to refresh a webpage at 6am any more — it emails you when the number changes.2025
- Dynamic Web ScraperPoint at the part of a page you want. Get it back as clean, structured data.2024