"What would you say if I told you there is an app on the ma-" -Jian Yang

Hotdog-not-Hotdog-ML

Repo for the ML model that classifies whether an image is a hotdog or not. This model is a CNN built with TensorFlow.

Motivation

We built this app for our Grade 12 Computer Science (ICS4U) class. This was a unique project as I was able to use Python for the course (which made the ML 100x easier :P). This is obviously a clone of SeeFood, a fictional app built by Jian Yang. The video is here:

Dataset

The dataset was provided by yasserhessein, Yashvardhan Jain, and tanliheng from kaggle, the links are: here, here, and here.

dataset
│
└───train
│   │
│   └───hot_dog
│   │   │   file111.png
│   │   │   file112.png
│   │   │   ...
│   │
│   └───not_hot_dog
│       │   file111.png
│       │   file112.png
│       │   ...
│
└───test
│   │
│   └───hot_dog
│   │   │   file111.png
│   │   │   file112.png
│   │   │   ...
│   │
│   └───not_hot_dog
│       │   file111.png
│       │   file112.png
│       │   ...
│

• 2866 images each of hotdogs and not hotdogs in the train directory
• 714 images each of hotdogs and not hotdogs in the test directory

Credits

Connor Wilson built the frontend with Java and I built the ML model with Python.