Duration: 24 hours
Build a creative computer vision application using:
- Focoos platform for model training
- ONNX export format with quantization library
- Arduino Nicla Vision as deployment target
- Zant for flashing your model
Use Case: Any computer vision application - we'll reward your creativity! 🎨
Submit via Pull Request to this repository
.inoArduino sketch.onnxmodel filemodel_info.jsonmetadataREADME.mddocumentation- anything relevant ( web pages, mobileapp, images, examples ... )
- 5 slides presentation deck (PDF) template here
- ✅ All submissions reviewed
- 🎤 Top 5 teams selected for live demo
- 🏆 3 winning teams announced
Most innovative and original application concept
Best code quality, documentation, and software engineering practices
Most technically impressive or groundbreaking implementation
As you build your solution, here are some aspects worth considering:
On Data & Models
- How did you choose your dataset? Edge AI has unique constraints that might influence what data works best
- Why did you pick that specific model architecture? Sometimes smaller is better... 😉
- Have you thought about the trade-offs between accuracy and efficiency for your particular use case?
On Performance & Deployment
- Getting your model running on actual hardware is just the beginning - does it run fast enough for what you're trying to do?
- A surveillance system might need different framerates than a periodic quality checker
- What if some of the all computation didn't happen on the device? Pre-processing, post-processing, or complementary algorithms could live elsewhere
On Integration & User Experience
- The Nicla Vision can communicate with other systems - how reliable is your connection?
- Could you visualize your system's output in a way that anyone could understand, not just developers?
- Have you explored what other sensors the Nicla Vision has beyond the camera? IMU, microphone, environmental sensors...
On Going Further
- What if you ran more than one model? The hardware might surprise you with what it can handle
- Think about the full pipeline: what happens before and after your model runs?
- How would someone without technical knowledge interact with your system?
- Focoos Platform: Model training & export
- Zant Tool: Model deployment to hardware
- Arduino Nicla Vision: Edge AI hardware platform
Good luck, and remember: creativity counts! 🎉