Update app.py

app.py

@@ -1,23 +1,8 @@
-import os
 import gradio as gr
 import torch
-import cv2
-import numpy as np
 from PIL import Image
 from transformers import AutoImageProcessor, AutoModelForImageClassification
 
-# Download cascade classifier if not present.
-cascade_filename = "haarcascade_frontalface_default.xml"
-if not os.path.exists(cascade_filename):
-    os.system(
-        "wget -O "
-        f"{cascade_filename} "
-        "https://raw.githubusercontent.com/opencv/opencv/master/data/haarcascades/haarcascade_frontalface_default.xml"
-    )
-
-# Load OpenCV's Haar Cascade face detector.
-face_cascade = cv2.CascadeClassifier(cascade_filename)
-
 # Load the Hugging Face model and processor for deepfake detection.
 processor = AutoImageProcessor.from_pretrained("Smogy/SMOGY-Ai-images-detector")
 model = AutoModelForImageClassification.from_pretrained("Smogy/SMOGY-Ai-images-detector")
@@ -32,51 +17,13 @@ def detect_deepfake(image: Image.Image) -> str:
     conf = probs[0, idx].item()
     return f"The image is {label} with confidence {conf:.2f}"
 
-def swap_faces(source: Image.Image, target: Image.Image) -> Image.Image:
-    # Convert to OpenCV BGR
-    src = cv2.cvtColor(np.array(source), cv2.COLOR_RGB2BGR)
-    tgt = cv2.cvtColor(np.array(target), cv2.COLOR_RGB2BGR)
-
-    # Convert to grayscale for detection
-    gray_src = cv2.cvtColor(src, cv2.COLOR_BGR2GRAY)
-    gray_tgt = cv2.cvtColor(tgt, cv2.COLOR_BGR2GRAY)
-
-    # Detect faces (x, y, w, h)
-    src_faces = face_cascade.detectMultiScale(gray_src, scaleFactor=1.1, minNeighbors=5)
-    tgt_faces = face_cascade.detectMultiScale(gray_tgt, scaleFactor=1.1, minNeighbors=5)
-
-    if len(src_faces) == 0 or len(tgt_faces) == 0:
-        return None
-
-    # Take the first detected face
-    x_s, y_s, w_s, h_s = src_faces[0]
-    x_t, y_t, w_t, h_t = tgt_faces[0]
-
-    # Extract and resize
-    face_src = src[y_s:y_s+h_s, x_s:x_s+w_s]
-    face_resized = cv2.resize(face_src, (w_t, h_t))
-
-    # Swap
-    tgt[y_t:y_t+h_t, x_t:x_t+w_t] = face_resized
-
-    # Convert back to PIL
-    return Image.fromarray(cv2.cvtColor(tgt, cv2.COLOR_BGR2RGB))
-
 # Build Gradio interface
 with gr.Blocks() as demo:
-    gr.Markdown("# Deepfake Detection & Face Swap App")
-
-    with gr.Tab("Detection"):
-        gr.Markdown("### Upload an image to detect deepfake content.")
-        img_in = gr.Image(type="pil", label="Upload Image")
-        txt_out = gr.Textbox(label="Result")
-        gr.Button("Detect").click(fn=detect_deepfake, inputs=img_in, outputs=txt_out)
-
-    with gr.Tab("Face Swap"):
-        gr.Markdown("### Upload source and target images for face swap.")
-        src_img = gr.Image(type="pil", label="Source Face")
-        tgt_img = gr.Image(type="pil", label="Target Image")
-        out_img = gr.Image(type="pil", label="Swapped Image")
-        gr.Button("Swap Faces").click(fn=swap_faces, inputs=[src_img, tgt_img], outputs=out_img)
-
-demo.launch()
+    gr.Markdown("# Deepfake Detection App")
+    gr.Markdown("### Upload an image to detect deepfake content.")
+    img_in = gr.Image(type="pil", label="Upload Image")
+    txt_out = gr.Textbox(label="Result")
+    gr.Button("Detect").click(fn=detect_deepfake, inputs=img_in, outputs=txt_out)
+
+if __name__ == "__main__":
+    demo.launch()