app.py

import gradio as gr
import boto3
from botocore.exceptions import BotoCoreError, ClientError
from PIL import Image
import numpy as np
import io
import pyttsx3
import time
import os
import subprocess


aws_key_id = os.environ['aws_access_key_id']
aws_secret = os.environ['aws_secret_access_key']





# Initialize AWS Rekognition client
try:
    client = boto3.client('rekognition', 
                      region_name='us-east-1',
                      aws_access_key_id= aws_key_id, 
                      aws_secret_access_key = aws_secret)
except (BotoCoreError, ClientError) as error:
    print('Error: ', error)

def speak_text(text):
    # Specify the text to be converted to speech
    text = "Hello, this is an example of using the built-in `espeak-ng` command for text-to-speech conversion on Linux."
    
    # Use the `espeak-ng` command to convert the text to speech and play it
    os.system(f"espeak-ng '{text}'")





def recognize_emotions(image):
    """
    This function takes an image as input, and returns the emotion with the highest confidence level in the face using AWS Rekognition
    """
    # Convert the NumPy array to PIL image
    pil_image = Image.fromarray(np.uint8(image))
    
    # Convert the PIL image to bytes
    with io.BytesIO() as output:
        pil_image.save(output, format="JPEG")
        contents = output.getvalue()
    
    # Perform detection on the image using AWS Rekognition
    response = client.detect_faces(
        Image={
            'Bytes': contents
        },
        Attributes=['ALL']
    )
    
    # If no faces are detected, return None
    if not response['FaceDetails']:
        return None
    
    # Extract the emotions detected in the face
    emotions = response['FaceDetails'][0]['Emotions']
    
    # Find the emotion with the highest confidence level
    max_confidence = 0
    max_emotion = ''
    for emotion in emotions:
        if emotion['Confidence'] > max_confidence:
            max_confidence = emotion['Confidence']
            max_emotion = emotion['Type']


    speak_text(f'This person is {max_emotion}')
    #espeak(f'This person is {max_emotion}')
    
    # Return the emotion with the highest confidence level as a string
    return str(max_emotion)

# Create Gradio interface
iface = gr.Interface(recognize_emotions, 
                     inputs=gr.Image(source="webcam", streaming=True),
                     outputs="text",
                     title="How does this person feel?",
                     description="Helping you understand what others think")

# Launch the interface
iface.launch()