| import pathlib | |
| import platform | |
| import gradio as gr | |
| from fastai.vision.all import load_learner | |
| from PIL import Image | |
| if platform.system() != 'Windows': | |
| pathlib.WindowsPath = pathlib.PosixPath | |
| EXPORT_PATH = "export.pkl" | |
| learn_inf = load_learner(EXPORT_PATH) | |
| def classify_image(img): | |
| """Classifies an image according to three categories: dung beetle, elephant, or dolphin. | |
| Args: | |
| img (any): Any image will be converted to expected type. | |
| Returns: | |
| _type_: Probabilies according to the three types. | |
| """ | |
| # Convert the image to a format the model expects | |
| img = Image.fromarray(img.astype('uint8'), 'RGB') | |
| # Make a prediction | |
| pred_class, pred_idx, probs = learn_inf.predict(img) | |
| # Return the result | |
| return {learn_inf.dls.vocab[i]: float(probs[i]) for i in range(len(learn_inf.dls.vocab))} | |
| demo = gr.Interface( | |
| title = "A dung beetle / dolphin / elephant image classifier", | |
| fn=classify_image, | |
| inputs = gr.Image( | |
| label = 'Upload an image of a dung beetle, a dolphin, or an elephant!'), | |
| outputs="label") | |
| if __name__ == "__main__": | |
| demo.launch(share=True) |