app.py

import os
import numpy as np
import cv2
import pickle
from flask import Flask, render_template, request
from tensorflow.keras.applications.mobilenet_v2 import preprocess_input
from tensorflow.keras.applications import MobileNetV2
from tensorflow.keras.models import Model
from tensorflow.keras.layers import GlobalAveragePooling2D
from sklearn.metrics.pairwise import cosine_similarity
from werkzeug.utils import secure_filename

app = Flask(__name__)
UPLOAD_FOLDER = 'static/uploads'
app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER

# 📌 Model
base_model = MobileNetV2(include_top=False, input_shape=(224, 224, 3), weights='imagenet')
x = GlobalAveragePooling2D()(base_model.output)
embed_model = Model(inputs=base_model.input, outputs=x)

# 🔁 Embedding verileri
with open("embeddings.pkl", "rb") as f:
    embeddings, labels = pickle.load(f)

# 🔍 Tahmin fonksiyonu
def predict_class(image_path):
    img = cv2.imread(image_path)
    if img is None:
        return "Görsel okunamadı", 0.0
    img = cv2.resize(img, (224, 224))
    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    img = preprocess_input(img)
    img = np.expand_dims(img, axis=0)

    test_emb = embed_model.predict(img, verbose=0)
    sims = cosine_similarity(test_emb, embeddings)[0]
    best_idx = np.argmax(sims)
    best_label = labels[best_idx]
    confidence = sims[best_idx]

    return best_label, confidence

# 🌐 Ana route
@app.route('/', methods=['GET', 'POST'])
def index():
    if request.method == 'POST':
        file = request.files['file']
        if file:
            filename = secure_filename(file.filename)
            filepath = os.path.join(app.config['UPLOAD_FOLDER'], filename)
            file.save(filepath)

            prediction, prob = predict_class(filepath)

            return render_template('index.html', filename=filename, prediction=prediction, prob=prob)

    return render_template('index.html')

# 🔥 Uygulama başlat
if __name__ == '__main__':
    app.run(debug=True, use_reloader=False)