This is my own version of the SeeFood app from the Silicon Valley TV show by HBO. It identifies if the image contains a hotdog or not. It uses pytorch and torchvision to to perform transfer learning on a pretrained resnet model and then uses the model to predict if the image contains a hotdog or not.
It's written using Jupyter Notebooks so feel free to checkout Seefood.ipynb and run it yourself. It will also show you some demonstrations of the model in action.
-
The source of the idea is from https://www.youtube.com/watch?v=vIci3C4JkL0
-
And you can find the data set on Kaggle: https://www.kaggle.com/datasets/dansbecker/hot-dog-not-hot-dog