ADD: example images

app.py

@@ -3,8 +3,9 @@ import cv2
 from PIL import Image
 import torch
 import numpy as np
+import os
 
-from transformers import AutoImageProcessor, AutoProcessor, AutoModel, CLIPVisionModel
+from transformers import AutoProcessor, CLIPVisionModel
 from detection import detect_image, detect_video
 from model import LinearClassifier
 
@@ -65,6 +66,14 @@ def change_input(input_type):
     else:
         return None
 
+def determine_model_type(image_path):
+    if "facial" in image_path.lower():
+        return "Facial"
+    elif "general" in image_path.lower():
+        return "General"
+    else:
+        return "Facial"  # 기본값
+
 
 def process_input(input_type, model_type, image, video):
     detection_type = "facial" if model_type == "Facial" else "general"
@@ -77,6 +86,18 @@ def process_input(input_type, model_type, image, video):
         return None, None
 
 
+def process_example(image_path):
+    model_type = determine_model_type(image_path)
+    return Image.open(image_path), model_type
+
+
+example_images = [
+    "examples/fake/facial.jpg",
+    "examples/fake/general.jpg",
+    "examples/real/facial.jpg",
+    "examples/real/general.jpg",
+]
+
 with gr.Blocks() as demo:
   
     gr.Markdown("## Deepfake Detection : Facial / General")
@@ -92,6 +113,15 @@ with gr.Blocks() as demo:
 
     pred_score_output = gr.Textbox(label="Prediction Score")
     attn_map_output = gr.Image(type="pil", label="Attention Map")
+
+    # Example Images 추가
+    gr.Examples(
+        examples=example_images,
+        inputs=[image_input],
+        outputs=[image_input, model_type],
+        fn=process_example,
+        cache_examples=False
+    )
   
     input_type.change(fn=change_input, inputs=[input_type], outputs=[image_input, video_input])