1. Overview
Our team made significant progress across all of the project components, with our main focus being preparing for the midterm demo. We refined our computer vision model, made significant progress on our motorized camera slider subsystem, and refined the user interface and user experience for our mobile application. Our challenges include difficulties of using our existing cameras and reliability issues with our cloud data pipeline.
2. Key Achievements
Hardware and Embedded Systems
- We made progress on the motorized camera slider subsystem, constructing the slider and powering it with a Nema 17 stepper motor integrated with a Raspberry Pi
- Completed calibration, optimized translation speed and stepping requirements for full traversal
- Integrated ring light system to ensure consistent lighting throughout scan
Computer Vision
- Refined the training pipeline for the object detection model, maximizing performance for demo
- Completed preliminary training, working on optimizing hyperparameters and training sequence and maximizing accuracy
Mobile App Development
- Improved user interface and user experience of our mobile app
- Implemented several UI improvements to ensure smooth and visually coherent experience for demo purposes
3. Next Steps
Hardware and Embedded Systems
- Add stable baseplate mount to slider to improve stability
- Integrate camera system with slider and complete full system testing
Computer Vision
- Continue expanding dataset and tuning hyperparameters to optimize accuracy
- Integrate model with pipeline for fridge item detection
Mobile App Development
- Implement recommendation system
- Continue backend integration to improve data storage and transmission
4. Outlook
Our team is on track despite minor delays, and will present functioning demos showcasing hardware control as well as user interface. We will continue working on feature development and tighter integration between our hardware and software systems.