Hello everyone!
I want to share a project I created to detect vehicle license plates using a Home Assistant camera through the Gemini 3.5 Flash API. It is extremely fast, highly intelligent (reads plates perfectly even at difficult angles), and does not require heavy local add-ons. When a recognized car arrives, it automatically triggers the gate switch and sends personalized notifications (TTS via Alexa and a pop-up alert on an LG webOS TV).
Since this setup requires a small helper script, installation is done in 3 simple steps:
1. The Python Script
Create a file at /config/www/tmp/analisar_matricula.py and paste the following code (don't forget to insert your own Gemini API Key in the configurations section):
Python
import sys
import requests
import base64
# ==============================================================================
# CONFIGURATIONS (The user must change these details)
# ==============================================================================
API_KEY = "PASTE_YOUR_GEMINI_API_KEY_HERE"
MODELO = "gemini-3.5-flash"
IMAGE_PATH = "/config/www/tmp/check_plate.jpg"
# ==============================================================================
def encode_image(image_path):
with open(image_path, "rb") as image_file:
return base64.b64encode(image_file.read()).decode('utf-8')
def analise_matricula():
try:
base64_image = encode_image(IMAGE_PATH)
except Exception as e:
# If the image doesn't exist at the moment, warn and exit safely
print(f"Error: Image not found at {IMAGE_PATH}")
sys.exit(1)
url = f"https://generativelanguage.googleapis.com/v1beta/models/{MODELO}:generateContent?key={API_KEY}"
headers = {
"Content-Type": "application/json"
}
payload = {
"contents": [{
"parts": [
{
"text": "Read the license plate in this image. Respond ONLY with the license plate text (e.g., AA00BB). If there is no license plate, respond 'NADA'."
},
{
"inline_data": {
"mime_type": "image/jpeg",
"data": base64_image
}
}
]
}]
}
try:
response = requests.post(url, headers=headers, json=payload, timeout=8)
if response.status_code == 200:
resultado = response.json()
# Safe extraction of the license plate text returned by Gemini
texto = resultado['candidates'][0]['content']['parts'][0]['text']
print(texto.strip())
else:
# Show the error if the Gemini API fails
print(f"API Error: {response.status_code}")
except Exception as e:
print(f"Connection error")
if __name__ == "__main__":
analise_matricula()
2. The Shell Command
Add the following lines to your configuration.yaml file, then go to Developer Tools -> YAML and Reload Shell Commands (or restart Home Assistant):
YAML
# Command that executes the license plate reading Python script
shell_command:
analisar_matricula: "python3 /config/www/tmp/analisar_matricula.py"
3. The Blueprint
You can import the Blueprint directly by copying the URL below and adding it to your Blueprints dashboard:
- Import Link:
https://gist.githubusercontent.com/Rveiga86/cbe638cbfa9f7987af72f59b1aedb3ad/raw/d05766668dbe4c74d45fe3850fa465870f1dcb1e/gemini_matriculas.yaml
Once imported, simply create a new automation from this Blueprint and fill in the required fields (Sensors, Gate Switch, License Plates, and Car Names) directly through the Visual User Interface (UI). No further YAML editing required!
I hope you find this useful. Let me know if you have any questions!
