Inspiration
We originally wanted to create an IoT project that would have a practical purpose. we narrowed it down to a camera activated lock. The intention was for it to recognize a face and unlock accordingly.
What it does
The webcam is attached to the top of the door where it captures footage and is processed on a raspberry pi with TensorFlow lite and OpenCV running to determine what objects are in the frame. When a keyboard is detected it moves a servo which unlocks the door
How we built it
The code was based on a general example of object detection with TensorFlow lite. From there it was adapted to trigger a servo motor. Most of the components were taped or hot-glued down.
Challenges we ran into
It took about 8 hours to get TensorFlow working on a raspberry pi and another 4 or so hours to successfully control the servo motor. The final challenge was figuring out how to position and mount the servo so it would be able to unlock the door with very little torque.
Accomplishments that I'm proud of
Managing to get TensorFlow to work on a raspberry pi and having a satisfying product that works most of the time
What I learned
Python as well as servo control on a raspberry pi
What's next for Alternate door unlock
Reliability and better recognition so individual faces can be recognized
Built With
- python
- tensorflowlite
Log in or sign up for Devpost to join the conversation.