Video demo: or


Food waste is a pressing issue that not only has financial ramifications, but also vast environmental effects. About 33% (1.3 billion tonnes) of global fresh produce is thrown away annually due to their quality not meeting standards. This food waste is estimated to cost the US alone $161B in revenue anually. Additionally, the carbon footprint caused by the transport and disposal of produce creates a significant net negative.

What it does

Fresh2Go provides an unique perspective to this problem. Utilizing Computer Vision and ML, we have created an algorithm that can detect the freshness of a produce, which can be used in a commerce platform to distribute it.

Our system connects buyers and suppliers to eliminate food waste. Suppliers can create a listing of their shipment and upload pictures of cargo, which will then be assigned a freshness rating from our server. Buyers can access shipment information, such as the location and contents of the shipment, and also specify their desired freshness and price range. If a shipment does not meet the required freshness of a buyer, other buyers can have the opportunity to bid on the shipment.

Fresh2Go aims to provide a new B2B market that benefits both parties and eliminates food waste.

How we built it

We used a model trained on thousands of images of produce using Keras as our main AI library. This model ran on a flask server that would take the picture uploads and respond with the type of produce in the picture as well as whether it was fresh enough to eat. One of our team members focused on the design and implementation of our front end in HTML5, one set up the flask server and the last implemented the model and prediction engine.

Challenges we ran into

None of us have much experience with UI design and front end work so getting the front end together was a challenge

Accomplishments that we’re proud of

All of us are relatively new in the realms of AI and ML so it was a scramble to learn everything we could about modern Python AI libraries. In spite of that, we were able to implement a relatively accurate and functional model

What we learned

  • Modern Python AI libraries like Keras and Tensorflow
  • Server side scripting in Flask
  • Collaboration gets the job done
  • Python is a versatile language
  • A good dataset is truly more valuable than oil

What's next for Fresh2Go

  • Determine how long fresh food will last and how long food has been rotten
  • Show nutrition info for food as well as health facts
  • Identify freshness of cooked foods and not just produce
  • Become a platform to donate food to those in need

Built With

Share this project: