Misfit Munch

Connecting "imperfect" produce from farmers with communities is a perfect solution to reduce modern food waste and food insecurity.

Comment

Inspiration

Our initiative confronts the urgent issue of food waste with a dual approach involving consumers and farmers. We implement a unique and creative solution using AI, computer vision, and drone technology.

“Ugly” vegetables are a major cause of food waste. Every year, over 20% of wasted fruit and vegetables were attributed to “ugly” or “wonky” produce, that is six billion pounds. This produce was discarded for failing to meet retailer appearance standards

On the other hand, Low-income families have a high awareness of healthy diets, but cannot afford good-quality food according to this article. This creates a lot of health problems due to bad nutrition.

What it does

For Farmers: We provide a dedicated platform where farmers can market their 'imperfect produce,' which, while entirely edible, may bear minor cosmetic imperfections that render them unsellable in conventional markets.

For Users: Users simply photograph the contents of their fridge, and our sophisticated algorithm takes over. It crafts personalized recipes based on available ingredients and accesses data on surplus produce from local farms.

How we built it

Backend Development:

Flask: Our web application is built on the Flask framework, providing us the flexibility and tools necessary to create a dynamic and interactive user experience. SQLite: Serving as our database, SQLite efficiently manages data and ensures smooth operations even as the user base grows.

Data Pre-population:

Faker: To simulate a realistic environment for testing, we utilized Faker to pre-populate our database with synthetic yet believable data.

Image Analysis:

Google Cloud Vision API (vision_v1): Post-segmentation, each segmented object is analyzed using the Google Cloud Vision API to identify and catalog the items present.

DeepLabV3: To dissect images of refrigerated items into distinct objects, we employed the DeepLabV3 model. This segmentation stage is crucial for accurate item identification.

Drone Prototyping:

Drone delivery is an eco-friendly and swift technological method that eliminates the inconvenience of farmers relying on other transportation while also actively contributing significantly to the sustainability of the environment.

CAD Modeling: Conceptualizing and iterating our drone design was made feasible and efficient through Computer-Aided Design (CAD) modeling.

Autodesk Fusion 360: For a precise and detailed design of our drone prototype, we used Autodesk Fusion 360, taking advantage of its comprehensive set of tools for 3D design, engineering, and simulation.

Challenges we ran into

  1. One of the challenges we faced was breaking down our project ideas into smaller, manageable tasks and then figuring out which tasks to work on first.
  2. When dealing with Google Cloud Computer Vision, it only does image recognition of one whole image and does not break it down. We discussed and came up with a creative solution where we used a different ML model to separate the original image and send each segment to Google Cloud.
  3. Our API key says "too many requests" at the end when we are trying to record a demo. We used too many requests in a short period of time while developing it

Accomplishments that we're proud of

  1. All of us came from different countries and we were able to match each other's energy very well. We have very different ideas which made us work so well together as we allow our thoughts to complement and amplify each other.
  2. Team member, Omari Emmanuel learned how to do HTML for the first time with everyone's help and successfully deployed the help web page in the main designed online platform for our project
  3. We researched deeply into a pressing issue that affects 25% of the US population, and we were able to come up with a comprehensive idea to mitigate that issue in a short amount of time.

What's next for Misfit Munch

  1. We want to continue working by trying to validate the market and see if there is a strong demand.
  2. Understand the local laws of drones and the viability of transporting goods and services using drones, figuring out how to engage with regulatory bodies.
  3. Secure funding by applying to grants, which also validates our ideas as well as giving us a head start in purchasing delivery drones
Share this project:

Updates