SeeFood

• 

  SeeFood

SeeFood

A concept brought to life. With technology increasingly making our lives easier, we wanted to create an application that not only simplifies cooking, but also aims to minimize waste. After years of eating only canned tuna and ramen noodles, it can be difficult to cook novel meals for oneself. This user-friendly app quickly takes in ingredients from the camera and quickly outputs relevant recipes that can be made with those ingredients.

We accomplished this by training on top of an existing convolution neural network (CNN) called Alexanet to develop our own image classifier specifically to identify common ingredients found in the everyday household. The CNN was compiled on NVIDIA DIGITS in the Caffe model format, then converted into CoreML format using "coremltools" to successfully integrate with the iOS system. To quicken up the training process we used 4 GTX1070Ti GPUs to increase the processing power of the computer. Here is an image:

Built With

• caffe
• coreml
• coremltools
• neural-network
• nvidia-digits
• python
• swift