Chinese and Western Food Classification

This repo contains code for running food classification by pytorch. It uses 208 chinese food classes and 101 western food classes(309-classes).
This repo is designed for those who want to start their projects of image classification.
It provides fast experiment setup and attempts to maximize the number of projects killed within the given time.
It includes a few Convolutional Neural Network modules.You can build your own dnn easily.

Requirements

• torch==0.4.1
• torchvision==0.2.1
• opencv
• Pillow

model

• mobilenetV2 (mobilenet_v2)

pre-trained model

you can download pretrain model in model_dir(CWFood_model.pth)

usage

Inference

python3 inference.py --image test.jpg

top-5:
Mapo_Tofu: 70.02458572387695%
Kung_Pao_Chicken: 5.765869095921516%
Spicy_Chicken: 5.33505454659462%
Home_style_sauteed_Tofu: 3.10797281563282%
Double_cooked_pork_slices: 2.423858270049095%

top-5:
pizza: 57.676124572753906%
garlic_bread: 8.819431811571121%
macaroni_and_cheese: 6.301581114530563%
paella: 4.138006269931793%
Pizza: 3.8569435477256775%

Experiments

There is integrated with the project using tensorboardX library which porved to be very useful as there is no official visualization library in pytorch. There is the learning curves for the food dataset experiment(top-1 acc: 69.66%).

Labels

train_data.txt
val_data.txt

References

1.https://sites.google.com/view/chinesefoodnet
2.http://www.vision.ee.ethz.ch/datasets_extra/food-101
3.https://pytorch.org
4.https://github.com/ericsun99/MobileNet-V2-Pytorch
5.https://github.com/frotms/image_classification_pytorch