Fresh Count | Devpost

Inspiration

We wanted an easy way to track pantry items, reduce food waste, and make inventory management hands-free.

What it does

Fresh Count captures images and voice notes of food, and translates voice notes to text via VOSK. On a front end web interface it displays foods with their quantity, expiration date, and stored location. Users can remove items from the web interface, upon removing all items that belong to a location the location itself will be deleted too.

How we built it

We integrated a smart camera and microphone with a backend system. Automatic image capture and voice recording feed data to the backend, which processes and organizes inventory for the frontend.

Challenges we ran into

Ensuring accurate speech recognition and seamless synchronization between hardware and backend was tricky, especially for automatic capture without errors. Proper parsing after turning speech to text proved to be a challenge when only relying on VOSK and python.

Accomplishments that we're proud of

We created a system to track pantry items by showing the item to a camera and speaking out loud the name, expiration date, quantity of item, and location put. This is then successfully turned into text with the assistance of VOSK and utilizing gemini as a way to further interpret the text from speech.

What we learned

We gained experience utilizing the python FastAPI library to connect a Raspberry Pi hardware with a backend and then provide the data to the frontend. We learned how to utilize VOSK to turn Speech to Text and then using a Gemini 2.5 Flash Lite to parse the text into a format that we can enter into a database.

What's next for Fresh Count

Future improvements include expanding recognition accuracy, adding notifications for expiring items, and integrating smart home features for even smarter pantry management.