---
title: FoodNet
emoji: 🍔
colorFrom: purple
colorTo: purple
sdk: streamlit
app_file: foodnet.py

---

# 24-679 FoodNet Project 

## Authors

David Chuan-En Lin: chuanenl@cs.cmu.edu

Mitch Fogelson: mfogelso@andrew.cmu.edu

Sunny Yang: yundiy@andrew.cmu.edu

Shihao Xu: shihaoxu@andrew.cmu.edu

## TODO

### Must Have

1. Cooking method (How to do this?) (TBD)
2. Ingredients -> Recipe (Recipe Querey?) (Mitch)
3. Cuisine Meta Data (Where to get) (TBD)
4. Deployment on the cloud -> (David)

### Like to have

1. Images related -> 

  * [Google Image Search API](https://pypi.org/project/Google-Images-Search/)
  * [OpenAI Clip](https://openai.com/api/)

2. User Studies 

### Moonshot

1. Recipe Masking Prediction 
2. 

## Description 

We wanted to help students and households in the Pittsburgh to reduce their food waste. We developed a model that suggests recipes based on current leftovers availible. 

* Model -> Facebook's [FastText](https://radimrehurek.com/gensim/models/fasttext.html) 
* Dataset -> [Simplified 1M+ Recipes](https://github.com/schmidtdominik/RecipeNet)
    * [Dominick Schmidt Blog](https://dominikschmidt.xyz/simplified-recipes-1M/#dataset-sources)

## Try WebApp

https://huggingface.co/spaces/chuanenlin/foodnet

## Quick Start

1. Clone repository 

```
git clone git@github.com:chuanenlin/foodnet.git
```

2. Move into repository 

```
cd foodnet
```

(**Optional** Create conda environment)

3. Install gdown

```
pip install gdown
```

4. Download models

```
gdown https://drive.google.com/drive/folders/1LlQpd45E71dSfC8FgvIhJjQjqxnlBC9j -O ./models --folder
```

5. Download datasets (Optional)

```
gdown https://drive.google.com/drive/folders/18aA3BFKqzkqNz5L4N5vN6bFnp8Ch2CQV -O ./data --folder
```

6. Install Dependencies

```
pip install -r requirements.txt
```

7. Run code

```
streamlit run foodnet.py
```

## Args

Train new model

```
streamlit run foodnet.py -d/--dataset ['/PATH/TO/DATASET'] -t/--train True 
```

Load alternative model

```
streamlit run foodnet.py --model ['/PATH/TO/MODEL'] 
```

## Requirements 

* python>=3.6
* gensim>=4.0.x
* streamlit
* gdown
* nltk
* pickle
* matplotlib

## References

TODO