app.py

import uvicorn
from fastapi import FastAPI, File, UploadFile
from PIL import Image
import numpy as np
import tensorflow as tf
import io

app = FastAPI()

# --- 1. CONFIGURATION ---
MODEL_PATH = "Orangemodel.tflite"

# ⚠️ IMPORTANT: You must list your classes here in alphabetical order!
# (TFLite does not store these names, so you must type them manually)
CLASS_NAMES = ['Citrus canker', 'Citrus greening', 'Citrus mealybugs', 'Die back', 'Foliage damaged', 'Healthy leaf', 'Powdery mildew', 'Shot hole', 'Spiny whitefly', 'Yellow dragon', 'Yellow leaves'] 

# --- 2. LOAD TFLITE MODEL ---
# We use the Interpreter, which is like a lightweight player for the model
interpreter = tf.lite.Interpreter(model_path=MODEL_PATH)
interpreter.allocate_tensors()

# Get input and output details so we know how to talk to the model
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()

@app.get("/")
def home():
    return {"message": "Orange Disease Detection API (TFLite Version) is Running!"}

@app.post("/predict")
async def predict(file: UploadFile = File(...)):
    # --- 3. PREPROCESS IMAGE ---
    # Read the file uploaded by the user
    image_data = await file.read()
    image = Image.open(io.BytesIO(image_data)).convert("RGB")

    # Resize to 224x224 (The size MobileNetV2 expects)
    image = image.resize((224, 224))

    # Convert to numpy array
    input_data = np.array(image)

    # Normalize the image (0 to 1) just like in training
    input_data = input_data / 255.0

    # Add the batch dimension (1, 224, 224, 3)
    input_data = np.expand_dims(input_data, axis=0)

    # TFLite requires FLOAT32 data type
    input_data = input_data.astype(np.float32)

    # --- 4. RUN INFERENCE ---
    # Tell the interpreter to look at our image
    interpreter.set_tensor(input_details[0]['index'], input_data)
    
    # Run the model
    interpreter.invoke()

    # Get the results
    output_data = interpreter.get_tensor(output_details[0]['index'])
    
    # --- 5. PROCESS RESULTS ---
    # Find the highest score
    prediction = np.argmax(output_data[0])
    confidence = float(np.max(output_data[0]))
    
    predicted_class = CLASS_NAMES[prediction]

    return {
        "class": predicted_class,
        "confidence": confidence
    }