• 사진 안에 있는 음식이 어떤 음식인지 인식할 수 있는 인공지능 모델 개발
• 사진 안에 있는 식재료가 어떤 식재료인지 인식할 수 있는 인공지능 모델 개발

You can start on any computer that can learn deep learning. If you want to learn fast, use GPU-workstation.

python version == 3.6.9

torch~=1.9.1
torchvision~=0.10.1
Pillow~=7.0.0
natsort~=7.0.1
sklearn~=0.0
scikit-learn~=0.24.2
tqdm~=4.42.0
numpy~=1.18.1
tensorflow~=2.2.0
tensorflow-gpu~=2.2.0
tensorboard~=2.7.0
matplotlib~=3.1.2

pip install -r requirements.txt

• If you only try inference,

torch~=1.9.1
torchvision~=0.10.1
Pillow~=7.0.0
natsort~=7.0.1
numpy~=1.18.1

If you respond to GoogleForms, we will share the download link within a few days. Currently, the shared checkpoint is ResNet152.

• Resquest JIT Traced Checkpoint
• Request Pytorch Checkpoint

• Use TorchScript

test_model = torch.jit.load('./jit_traced_torch_model_name.pt', map_location='cpu')
sample_data = torch.randn(1, 3, 512, 512) # (1, channel, width, height)
out_data = test_model(sample_data)

• pytorch

python inference.py --config configure_name.json --image image_name.jpg --label labels.txt

1. You need to create a configuration first.
2. Then execute the following command:

python train.py --configuration configuration_name.json

• KOREA AI HUB DATASET - 한국 음식 이미지 데이터셋
• Fruit and Vegetable Image Dataset - 과일과 채소 데이터셋
• Vegetable Image Dataset - 채소 데이터셋
• Private Dataset (INGD_V1, INGD_V2)

• Introduction to TorchScript - TorchScript 소개 및 튜토리얼
• Deep Java Library Pytorch - Pytorch용 Deep Java Library Engine Provider 소개 및 튜토리얼

• waverDeep - model architecture, setup train/test pipeline

This project is licensed under the MIT License - see the LICENSE.md file for details