An iOS app capable of detecting whether a photo contains a gyros or not
gyros-identifier ?️
In case you enter a shady restaurant, with suspicious prices and weird plate descriptions, not being able to verify that the food you ordered is a gyros… Fear no more! I made an app able to classify whether an image contains a gyros or not.
The idea stems from the clip of Jian Yang’s “Hotdog/Not Hotdog” app from the Silicon Valley series. Here’s the clip.
Editor note: The certainty whether this app contributes to society is controversial.
Boring tech details
The pretrained model used was the MLImageClassifier which was fine-tuned using the following custom dataset.
The dataset that was used was a mix of the Food-101 dataset from Kaggle and images of gyros that I downloaded from the internet, mostly from random sources which can be found here.
To split the dataset to Training and Test data I used the dataset_creator.sh bash script. Then I used the CreateML tool to train my model, and finally export it and add it to my XCode project.
Examples
| Example 1 | Example 2 |
|---|---|