Twitter Feed is a smart bird feeder that can seamlessly detect the presence of birds using a dedicated weight sensor and simultaneously take a picture which is displayed in an interactive web app. The app stores detection history, identifies bird detection trends (time of day and temperature) and shows alerts when your bird feed is running low.
- Twitter Feed can predict the breed of bird from the picture to various degrees of accuracy using a trained ML model and present data from previous days in dynamic graphs.
- Twitter Feed also tracks the weight of your bird feed at 5-second intervals and alerts you when food is below 50g.
- Twitter Feed keeps a history of the time of day and temperature (using a temperature sensor) of when each breed is detected, so you can have the best chance of seeing the bird in person.
- Twitter Feed uses a secure AWS authentication system where each user logs in with a unique username and password (with secure communication using AWS secret keys). Multiple users can be registered on one Twitter Feeder, and a single user can access multiple feeders in different locations by entering the Bird Feeder Serial ID when accessing the web app, allowing for scalability.
The promotional website for the product can be found here.
Here is the marketing video link.
Server code: To run the prediction server which identifies the bird breeds, run python3 prediction-server.py on ec2 instance with bird_model.py as the dependency in the bird-recognition-server/server folder. The main server which interfaces with the DynamoDB table, S3 bucket and React app is in the main-server folder. It also runs on an ec2 instance and can be run using node server.js
Web-app code: To run the React application, go into the react-app/amplifyapp/Frontend/src folder and run npm start.
Client code: In raspberry-pi, go into populating-dynamo-db and run python3 fake-upload.py to populate the DynamoDB table systematically with data that can be used for bird trends and historical information. The main.py raspberry pi code used in the demo can be found in raspberry-pi/demo
| Folder | File | Description |
|---|---|---|
| bird-recognition-server | bird_images | These are test images for testing the bird prediction |
| bird-recognition-server | local | Contains the pretrained bird identification model. This folder has code for testing the bird prediction on our local devices. |
| bird-recognition-server | server | Also contains the pretrained bird identification model. This folder has the code that was run on our bird prediction AWS server. |
| main-server | bird_images | Contains the code for the AWS server which fetches data from our DynamoDB table and interfaces with our react application. |
| react-app/amplifyapp | Frontend | Contains our React App code with AWS Amplify for authentication. |
| raspberry-pi | README.md | Contains the descriptions of the files in this folder. |
Main Dashboard showing birds detected and the breed predictions.
Identifying trends for different types of birds.
- Twitter_Table_New: stores all the data
- Twitter_User_Table: binds user_id and bird_feeder's serial number