|
|
--- |
|
|
license: mit |
|
|
--- |
|
|
A simple single label classification model, ResNet18, to predict whether the provided image is a cat or a dog. The model was created in Fast.ai |
|
|
and exported to ONNX using PyTorch's ONNX export capabilities. |
|
|
|
|
|
The source dataset is the OXFORD-IIIT PET. Omkar M Parkhi, Andrea Vedaldi, Andrew Zisserman and C. V. Jawahar |
|
|
We have created a 37 category pet dataset with roughly 200 images for each class. |
|
|
The images have a large variations in scale, pose and lighting. All images havean |
|
|
associated ground truth annotation of breed, head ROI, and pixel level trimap segmentation. |
|
|
|
|
|
The ONNX model can be used in other frameworks like Elixir's Axon. An example of converting the ONNX model into Axon can be found at: |
|
|
https://github.com/elixir-nx/axon/tree/main/notebooks/onnx_to_axon.livemd. |
|
|
|
