app.py

import gradio as gr
import cv2
import os
import boto3

aws_access_key_id = os.getenv('AWS_ACCESS_KEY_ID')
aws_secret_access_key = os.getenv('AWS_SECRET_ACCESS_KEY')

s3_client = boto3.client(
    's3',
    aws_access_key_id=aws_access_key_id,
    aws_secret_access_key=aws_secret_access_key,
    region_name='eu-central-1' 
)

def upload_to_s3(bucket_name, folder_name):
    image_paths = []
    for filename in os.listdir(folder_name):
        if filename.endswith('.png'):
            file_path = os.path.join(folder_name, filename)
            s3_client.upload_file(file_path, bucket_name, f"{folder_name}/{filename}")
            image_paths.append(file_path)
    return image_paths

def process_video(uploaded_video, name, surname, interval_ms):
    try:
        video_source = uploaded_video
        if video_source is None:
            return "No video file provided.", []

        folder_name = f"{name}_{surname}"
        os.makedirs(folder_name, exist_ok=True)

        # Video processing logic
        # Use video_source directly as it's a file path (string)
        temp_video_path = video_source
        face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
        vidcap = cv2.VideoCapture(temp_video_path)
        if not vidcap.isOpened():
            raise Exception("Failed to open video file.")

        fps = vidcap.get(cv2.CAP_PROP_FPS)
        frame_interval = int(fps * (interval_ms / 1000))
        frame_count = 0
        saved_image_count = 0
        success, image = vidcap.read()
        image_paths = []

        while success and saved_image_count < 86:
          if frame_count % frame_interval == 0:
            gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
            faces = face_cascade.detectMultiScale(gray, 1.2, 4)
            for (x, y, w, h) in faces:
                # Additional checks for face region validation
                aspect_ratio = w / h
                if aspect_ratio > 0.75 and aspect_ratio < 1.33 and w * h > 4000:  # Example thresholds
                    face = image[y:y+h, x:x+w]
                    face_resized = cv2.resize(face, (160, 160))
                    image_filename = os.path.join(folder_name, f"{name}_{surname}_{saved_image_count:04d}.png")
                    cv2.imwrite(image_filename, face_resized)
                    image_paths.append(image_filename)
                    saved_image_count += 1
                if saved_image_count >= 86:
                    break

          success, image = vidcap.read()
          frame_count += 1


        vidcap.release()

        bucket_name = 'newimagesupload00'
        uploaded_images = upload_to_s3(bucket_name, folder_name)

        return f"Saved and uploaded {saved_image_count} face images", uploaded_images

    except Exception as e:
        return f"An error occurred: {e}", []



# Gradio Interface
with gr.Blocks() as demo:
    gr.Markdown("### Video Uploader and Face Detector")
    gr.Markdown("Upload your own video to add your images to the dataset!")
    gr.Markdown("Make a short 10-15 seconds video of your front and side profiles, **slowly rotating your face**, with good lighting and visible face for best results.")
    with gr.Row():
        with gr.Column():
            video = gr.File(label="Upload Your Video!")
            
        with gr.Column():
            name = gr.Textbox(label="Name")
            surname = gr.Textbox(label="Surname")
            interval = gr.Number(label="Interval in milliseconds", value=100)
            submit_button = gr.Button("Submit")
            
        with gr.Column():
            gallery = gallery = gr.Gallery(
        label="Generated images", show_label=False, elem_id="gallery"
    , columns=[3], rows=[1], object_fit="contain", height="auto")

    submit_button.click(
        fn=process_video,
        inputs=[video, name, surname, interval],
        outputs=[gr.Text(label="Result"), gallery]
    )

css = """
body { font-family: Arial, sans-serif; }
"""
# Demo Launching
demo.launch()