app.py

# Gradio app interface to dog_breed_classifier model fine-tuned on kaggle. 
# This is the project from lesson 2 of the fastai Deep Learning course.
# 
# Reference:
# Kaggle: https://www.kaggle.com/code/mpfoley73/dog-breed-classification
# Dog Breed dataset: https://www.kaggle.com/datasets/khushikhushikhushi/dog-breed-image-dataset
# Tanishq blog: https://www.tanishq.ai/blog/posts/2021-11-16-gradio-huggingface.html
# Fastai: https://course.fast.ai/Lessons/lesson2.html
#

import gradio as gr
from fastai.vision.all import *
import skimage

learn = load_learner('dog_breed_classifier.pkl')

labels = learn.dls.vocab
def predict(img):
    img = PILImage.create(img)
    pred,pred_idx,probs = learn.predict(img)
    return {labels[i]: float(probs[i]) for i in range(len(labels))}

title = "Dog Breed Classifier"
description = "A dog breed classifier trained on the Dog Breed dataset with fastai. Created as a demo for Gradio and HuggingFace Spaces."
article="<p style='text-align: center'><a href='https://tmabraham.github.io/blog/gradio_hf_spaces_tutorial' target='_blank'>Blog post</a></p>"
examples = ['Chester 14.jpg']
interpretation='default'
enable_queue=True

# Construct a Gradio Interface object from the function (usually an ML model
# inference function), Gradio input components (the number should match the
# number of function parameters), and Gradio output components (the number 
# should match the number of values returned by the function).

gr.Interface(
    fn=predict,
    inputs=gr.inputs.Image(shape=(512, 512)),
    outputs=gr.outputs.Label(num_top_classes=3),
    title=title,
    description=description,
    article=article,
    examples=examples,
    interpretation=interpretation,
    enable_queue=enable_queue
).launch()