The E-Nose started as a final-year project: could cheap gas sensors and a Raspberry Pi tell when a pineapple is ripe?
Sensors drift. Plan for it.
Low-cost gas sensors change with temperature and humidity. Calibration runs and baseline correction mattered more than model choice.
[ E-NOSE PROTOTYPE ]
Small data, simple models
With a modest dataset, well-engineered features and a simple classifier beat anything deep. We reached 87% accuracy across ripeness stages.
From thesis to product
- Ruggedise the enclosure
- Add on-device recalibration
- Validate on other fruit
