Automated Herb Garden with a Raspberry Pi

The setup

A 60 cm × 20 cm indoor herb planter on a kitchen windowsill: basil (*Ocimum basilicum*), coriander (*Coriandrum sativum*), mint (*Mentha × piperita*), and flat-leaf parsley (*Petroselinum crispum*). A Raspberry Pi 4 (2 GB) with a Pi Camera Module v3 mounted to the cabinet above on a 3D-printed bracket.

Captures every 30 minutes. The pipeline runs P3b.0 (segmentation), P4.0 (chlorophyll), and P3a.0 (HSV colour).

Why automation matters for herbs

Herbs are fast-cycling crops. Basil in particular bolts quickly — the transition from vegetative to flowering takes 4–7 days once triggered by day-length or heat stress. By the time you see flower buds, you have probably already lost the best flavour in the leaves.

Sproutcast's colour pipeline detects the shift from deep green (high chlorophyll) toward yellow-green about 3–5 days before bolting is visible. That is enough lead time to pinch the growing tips and extend the harvest window.

The alert that saved the basil

In week eight of running, Sproutcast sent a turgor alert at 11 PM on a Thursday. The turgor proxy had dropped from 0.71 to 0.54 over two consecutive captures. I checked the plant — the soil was bone dry and the leaves were just starting to curl.

Without the alert I would have watered on Saturday. Two more days of stress would have damaged the roots.

Sharing the data

I export the weekly metric CSV (from the dashboard → Export data) and paste it into a simple Google Sheet to track trends over the season. The GBIF species key stored in Sproutcast links my data to the wider scientific literature on *Ocimum basilicum* phenology — useful if I ever want to compare against published growth curves.

What the Pi setup actually costs to run

The Pi 4 draws about 3.5 W under load. At a UK electricity rate of ~27p/kWh that is roughly £0.25 per month to run 24/7. The Pi Camera v3 cost £25. Total hardware: under £80.