app.py

from flask import Flask, request, jsonify
import tensorflow as tf
import numpy as np
import cv2

# Load your model
model = tf.keras.models.load_model('Realtime_CNN_Model.h5')

# Define your class names
classnames = ["angry", "disgust", "fear", "happy", "neutral", "sad", "surprise"]

app = Flask(__name__)

def detect_depression(sent_out):
    dc = 0  # negative emotion count
    hc = 0  # positive emotion count
    suc = 0 # surprise count

    for emotion in sent_out:
        if emotion in ['angry', 'disgust', 'sad', 'fear']:
            dc += 1
        elif emotion in ['happy', 'neutral']:
            hc += 1
        elif emotion == 'surprise':
            suc += 1

    pos_rate = round((hc / (dc + hc + suc)), 2) * 100
    neg_rate = round((dc / (dc + hc + suc)), 2) * 100

    response = ""
    depression_scale = round(neg_rate / 10, 1)
    
    if neg_rate <= 30.0:
        response = "*** MILD DEPRESSION ***\n"
        if neg_rate <= 5.0:
            response += "just chill and relax"
        elif neg_rate <= 10.0:
            response += f"Depression Scale Detected: {depression_scale}\nEverything is a-okay! You're probably cuddling a fluffy kitten right now."
        elif neg_rate <= 20.0:
            response += f"Depression Scale Detected: {depression_scale}\nYou are a bit frustrated and disappointed but easily distracted and cheered with little effort."
        elif neg_rate <= 30.0:
            response += f"Depression Scale Detected: {depression_scale}\nThings are bothering you but you're coping. You might be overtired and hungry."
    elif neg_rate <= 60.0:
        response = "*** MODERATE DEPRESSION ***\n"
        if neg_rate <= 40.0:
            response += f"Depression Scale Detected: {depression_scale}\nToday is slightly a bad day for you. Use self-care strategies."
        elif neg_rate <= 50.0:
            response += f"Depression Scale Detected: {depression_scale}\nYour mental health is starting to impact your everyday life."
        elif neg_rate <= 60.0:
            response += f"Depression Scale Detected: {depression_scale}\nYou are struggling with everyday tasks due to your mental health."
    else:
        response = "*** SEVERE DEPRESSION ***\n"
        if neg_rate <= 70.0:
            response += f"Depression Scale Detected: {depression_scale}\nYou are losing interest in previously enjoyable activities. Seek help."
        elif neg_rate <= 80.0:
            response += f"Depression Scale Detected: {depression_scale}\nYour mental health is affecting all parts of your life. You need urgent help."
        elif neg_rate <= 90.0:
            response += f"Depression Scale Detected: {depression_scale}\nYou are at a critical point. Seek help immediately."
        elif neg_rate <= 100.0:
            response += f"Depression Scale Detected: {depression_scale}\nSevere distress detected. Contact crisis line immediately."

    return response

@app.route('/predict', methods=['POST'])
def predict():
    file = request.files['file']
    img = cv2.imdecode(np.fromstring(file.read(), np.uint8), cv2.IMREAD_GRAYSCALE)
    img = cv2.resize(img, (48, 48))
    img = np.array(img).reshape(1, 48, 48, 1) / 255.0

    # Predict
    predict_x = model.predict(img)
    result = np.argmax(predict_x, axis=1)
    emotion = classnames[result[0]]

    # Apply depression detection logic
    sent_out = [emotion]  # This would typically be a batch of results
    depression_response = detect_depression(sent_out)

    return jsonify({'emotion': emotion, 'depression_analysis': depression_response})

if __name__ == '__main__':
    app.run(host='0.0.0.0', port=5000)