app.py

from fasthtml.common import *
from fasthtml.core import serve
from fastai.vision.all import *
import io
import base64
from PIL import Image as PILImageLib
from starlette.datastructures import UploadFile

# Load the model
learn = load_learner('model.pkl')
labels = learn.dls.vocab

# Create the app with nice styling
app, rt = fast_app(
    hdrs=(
        Style("""
        .prediction {
            padding: 10px;
            margin: 6px 0;
            border-radius: 6px;
            display: flex;
            justify-content: space-between;
        }
        .top-prediction {
            background-color: rgba(72, 187, 120, 0.15);
            font-weight: bold;
        }
        .preview-img {
            max-width: 300px;
            border-radius: 8px;
            margin: 15px 0;
            box-shadow: 0 4px 8px rgba(0,0,0,0.1);
        }
        .loader {
            border: 4px solid #f3f3f3;
            border-top: 4px solid #3498db;
            border-radius: 50%;
            width: 30px;
            height: 30px;
            animation: spin 2s linear infinite;
            margin: 20px auto;
            display: none;
        }
        .htmx-request .loader {
            display: block;
        }
        @keyframes spin {
            0% { transform: rotate(0deg); }
            100% { transform: rotate(360deg); }
        }
        .drop-container {
            border: 2px dashed #ccc;
            border-radius: 8px;
            padding: 20px;
            text-align: center;
            margin-bottom: 20px;
            transition: all 0.3s;
        }
        .drop-container:hover {
            border-color: #1095c1;
            background-color: rgba(16, 149, 193, 0.05);
        }
        """)
    )
)

@rt
def index():
    return Titled(
        "Image Classifier",
        Container(
            H1("Image Classifier"),
            P("Upload an image and our model will classify what's in it."),
            Div(
                Form(
                    Div(
                        Input(type="file", name="file", accept="image/*", required=True, id="file-upload"),
                        P("Drag and drop an image or click to select", cls="text-muted"),
                        cls="drop-container"
                    ),
                    Button("Classify", type="submit"),
                    hx_post="/classify",
                    hx_target="#result",
                    hx_encoding="multipart/form-data"
                ),
                Div(cls="loader"),
                Div(id="result")
            )
        )
    )

@rt("/classify")
async def post(file: UploadFile):
    # Read the file
    contents = await file.read()
    
    # Process with FastAI
    img = PILImage.create(io.BytesIO(contents))
    img = img.resize((512, 512))
    
    # Make prediction
    pred, pred_idx, probs = learn.predict(img)
    
    # Get top results
    results = [(labels[i], float(probs[i])) for i in range(len(labels))]
    results.sort(key=lambda x: x[1], reverse=True)
    top_results = results[:3]
    
    # Convert image for display
    buffered = io.BytesIO()
    pil_img = PILImageLib.fromarray(img.numpy().astype('uint8'))
    pil_img.save(buffered, format="JPEG")
    img_str = base64.b64encode(buffered.getvalue()).decode('utf-8')
    
    # Return the result
    return Card(
        H3("Classification Results"),
        Img(src=f"data:image/jpeg;base64,{img_str}", cls="preview-img"),
        Div(
            *[Div(
                Span(label),
                Span(f"{prob:.2%}"), 
                cls=f"prediction {'top-prediction' if i == 0 else ''}"
            ) for i, (label, prob) in enumerate(top_results)]
        ),
        Button("Classify Another", hx_get="/", hx_target="body")
    )

if __name__ == "__main__":
    # Start the server with Hugging Face port
    serve(port=7860)