Smart Scarecrow


We grow watermelon in a small garden near our house. The problem is that when birds are coming, they are eating and make holes in the watermelons. We tried to put a scarecrow but unfortunately it did not helped, the birds just ignore it and they keep destroy the watermelons.

What it does

We thought how to make the Scarecrow smart. every time when a bird appears near the watermelons, a loud alarm will be triggered and will scared the bird

How we built it

There is an Android tablet that take picture of the garden every minute It upload the picture to Azure Cognitive Services (Computer Vision) that analyze it and returns information about visual content found in an image If it found a bird in the picture then it trigger 30 seconds alarm.

On the site you can see all pictures and filter them by date and time. You can also apply tag filter bird/cat/dog and then it display only the pictures with those tags. Alerts log include all pictures with birds (for testing we include cat and dog).

The alarm trigger by switch using Azure IOT

We used: Azure App Service Azure Cosmos DB account Azure IOT Azure Cognitive Service Tracker app for Android

Challenges we ran into

Everything went well except Tablet & IOT integration which was complicated but we manage to solve it.

Accomplishments that we're proud of

We proud of the idea and the implementation on Microsoft Azure , we used Cloud Only solution for it. As we can do simple things more smarter – it’s great feeling , Thanks Azure services for that.

What we learned

We learned how to make the grow of the watermelons more smart and safety We also learned that Microsoft has great services and can serve any solution which you can think of.

What's next for Smart Scarecrow

Simple and cheap app for tablets The app will take picture every X time analyse the content it and will take action online according the rules we set. For example: Camera located in home will send SMS if there is fire or smoke in the picture. or close main water tap is case of water

Share this project: