xx025 / flyorbird

Use cifar10 dataset bird and airplane to train a model, then convert the model to onnx format, and deploy the model with gradio.

FlyOrBird

toy project resnet cifar10 classification pytorch torchvision onxx gradio

Use cifar10 dataset bird and airplane to train a model, then convert the model to onnx format, and deploy the model with gradio.

0. Install Requirements

conda create -n flyorbird python=3.10
pip install torch torchvision gradio numpy onnxruntime-gpu

Note: maybe is not enough, place by yourself.

1. Train Model

python train.py

2. Test Model

pretrained model weights can get for release

python test.py

3. Convert Model to ONNX

python convert2onnx.py

4. Deploy ONNX Model With Gradio

onxx model can get for release

python deploy.py

5. Test