Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -10,7 +10,9 @@ MODEL_IDENTIFIER = r"Ateeqq/ai-vs-human-image-detector"
|
|
10 |
DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
11 |
|
12 |
# --- Suppress specific warnings ---
|
|
|
13 |
warnings.filterwarnings("ignore", message="Possibly corrupt EXIF data.")
|
|
|
14 |
warnings.filterwarnings("ignore", message=".*You are using the default legacy behaviour.*")
|
15 |
|
16 |
|
@@ -26,9 +28,10 @@ try:
|
|
26 |
print("Model and processor loaded successfully.")
|
27 |
except Exception as e:
|
28 |
print(f"FATAL: Error loading model or processor: {e}")
|
|
|
29 |
raise gr.Error(f"Failed to load the model: {e}. Cannot start the application.") from e
|
30 |
|
31 |
-
# --- Prediction Function
|
32 |
def classify_image(image_pil):
|
33 |
"""
|
34 |
Classifies an image as AI-generated or Human-made.
|
@@ -41,21 +44,29 @@ def classify_image(image_pil):
|
|
41 |
confidence scores. Returns an empty dict if input is None.
|
42 |
"""
|
43 |
if image_pil is None:
|
|
|
44 |
print("Warning: No image provided.")
|
45 |
-
return {}
|
46 |
|
47 |
print("Processing image...")
|
48 |
try:
|
|
|
49 |
image = image_pil.convert("RGB")
|
|
|
|
|
50 |
inputs = processor(images=image, return_tensors="pt").to(DEVICE)
|
51 |
|
|
|
52 |
print("Running inference...")
|
53 |
with torch.no_grad():
|
54 |
outputs = model(**inputs)
|
55 |
logits = outputs.logits
|
56 |
|
57 |
-
probabilities
|
|
|
|
|
58 |
|
|
|
59 |
results = {}
|
60 |
for i, prob in enumerate(probabilities):
|
61 |
label = model.config.id2label[i]
|
@@ -67,22 +78,25 @@ def classify_image(image_pil):
|
|
67 |
except Exception as e:
|
68 |
print(f"Error during prediction: {e}")
|
69 |
# Return error in the format expected by gr.Label
|
70 |
-
|
|
|
71 |
|
72 |
# --- Define Example Images ---
|
73 |
example_dir = "examples"
|
74 |
example_images = []
|
75 |
-
if os.path.exists(example_dir):
|
76 |
for img_name in os.listdir(example_dir):
|
77 |
if img_name.lower().endswith(('.png', '.jpg', '.jpeg', '.webp')):
|
78 |
example_images.append(os.path.join(example_dir, img_name))
|
79 |
-
|
|
|
|
|
|
|
80 |
else:
|
81 |
print("No 'examples' directory found or it's empty. Examples will not be shown.")
|
82 |
|
83 |
|
84 |
# --- Custom CSS ---
|
85 |
-
# You can experiment with different CSS here
|
86 |
css = """
|
87 |
body { font-family: 'Inter', sans-serif; } /* Use a clean sans-serif font */
|
88 |
|
@@ -116,6 +130,14 @@ body { font-family: 'Inter', sans-serif; } /* Use a clean sans-serif font */
|
|
116 |
#prediction-label .label-name { font-weight: bold; font-size: 1.1em; }
|
117 |
#prediction-label .confidence { font-size: 1em; }
|
118 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
119 |
/* Style the examples section */
|
120 |
.gradio-container .examples-container { padding-top: 15px; }
|
121 |
.gradio-container .examples-header { font-size: 1.1em; font-weight: bold; margin-bottom: 10px; color: #34495e; }
|
@@ -143,8 +165,7 @@ body { font-family: 'Inter', sans-serif; } /* Use a clean sans-serif font */
|
|
143 |
"""
|
144 |
|
145 |
# --- Gradio Interface using Blocks and Theme ---
|
146 |
-
# Choose a theme: gr.themes.Soft(), gr.themes.Monochrome(), gr.themes.Glass()
|
147 |
-
# Or customize the default: gr.themes.Default().set(radius_size="sm", spacing_size="sm")
|
148 |
theme = gr.themes.Soft(
|
149 |
primary_hue="emerald", # Color scheme based on emerald green
|
150 |
secondary_hue="blue",
|
@@ -153,7 +174,7 @@ theme = gr.themes.Soft(
|
|
153 |
spacing_size=gr.themes.sizes.spacing_lg, # More spacing
|
154 |
).set(
|
155 |
# Further fine-tuning
|
156 |
-
body_background_fill="#
|
157 |
block_radius="12px",
|
158 |
)
|
159 |
|
@@ -179,7 +200,8 @@ with gr.Blocks(theme=theme, css=css) as iface:
|
|
179 |
submit_button = gr.Button("π Classify Image", variant="primary") # Make button prominent
|
180 |
|
181 |
with gr.Column(scale=1, min_width=300, elem_id="output-column"):
|
182 |
-
|
|
|
183 |
result_output = gr.Label(
|
184 |
num_top_classes=2,
|
185 |
label="Classification",
|
@@ -187,7 +209,7 @@ with gr.Blocks(theme=theme, css=css) as iface:
|
|
187 |
)
|
188 |
|
189 |
# Examples Section
|
190 |
-
if example_images: # Only show examples if they exist
|
191 |
gr.Examples(
|
192 |
examples=example_images,
|
193 |
inputs=image_input,
|
@@ -201,27 +223,24 @@ with gr.Blocks(theme=theme, css=css) as iface:
|
|
201 |
gr.Markdown(
|
202 |
"""
|
203 |
---
|
204 |
-
**How it Works:**
|
205 |
This application uses a fine-tuned [SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip) vision model
|
206 |
specifically trained to differentiate between images generated by Artificial Intelligence and those created by humans.
|
207 |
|
208 |
-
|
209 |
-
|
210 |
-
**Model:**
|
211 |
-
* You can find the model card here: <a href='https://huggingface.co/{model_id}' target='_blank'>{model_id}</a>
|
212 |
|
|
|
213 |
""".format(model_id=MODEL_IDENTIFIER),
|
214 |
elem_id="app-footer"
|
215 |
)
|
216 |
|
217 |
# Connect the button click or image change to the prediction function
|
218 |
-
|
219 |
-
|
220 |
-
image_input.change(fn=classify_image, inputs=image_input, outputs=result_output, api_name="
|
221 |
|
222 |
|
223 |
# --- Launch the App ---
|
224 |
if __name__ == "__main__":
|
225 |
print("Launching Gradio interface...")
|
226 |
-
iface.launch() # share=True
|
227 |
print("Gradio interface launched.")
|
|
|
10 |
DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
11 |
|
12 |
# --- Suppress specific warnings ---
|
13 |
+
# Suppress the specific PIL warning about potential decompression bombs
|
14 |
warnings.filterwarnings("ignore", message="Possibly corrupt EXIF data.")
|
15 |
+
# Suppress transformers warning about loading weights without specifying revision
|
16 |
warnings.filterwarnings("ignore", message=".*You are using the default legacy behaviour.*")
|
17 |
|
18 |
|
|
|
28 |
print("Model and processor loaded successfully.")
|
29 |
except Exception as e:
|
30 |
print(f"FATAL: Error loading model or processor: {e}")
|
31 |
+
# If the model fails to load, we raise an exception to stop the app
|
32 |
raise gr.Error(f"Failed to load the model: {e}. Cannot start the application.") from e
|
33 |
|
34 |
+
# --- Prediction Function ---
|
35 |
def classify_image(image_pil):
|
36 |
"""
|
37 |
Classifies an image as AI-generated or Human-made.
|
|
|
44 |
confidence scores. Returns an empty dict if input is None.
|
45 |
"""
|
46 |
if image_pil is None:
|
47 |
+
# Handle case where the user clears the image input
|
48 |
print("Warning: No image provided.")
|
49 |
+
return {} # Return empty dict, Gradio Label handles this
|
50 |
|
51 |
print("Processing image...")
|
52 |
try:
|
53 |
+
# Ensure image is RGB
|
54 |
image = image_pil.convert("RGB")
|
55 |
+
|
56 |
+
# Preprocess using the loaded processor
|
57 |
inputs = processor(images=image, return_tensors="pt").to(DEVICE)
|
58 |
|
59 |
+
# Perform inference
|
60 |
print("Running inference...")
|
61 |
with torch.no_grad():
|
62 |
outputs = model(**inputs)
|
63 |
logits = outputs.logits
|
64 |
|
65 |
+
# Get probabilities using softmax
|
66 |
+
# outputs.logits is shape [1, num_labels], softmax over the last dim
|
67 |
+
probabilities = torch.softmax(logits, dim=-1)[0] # Get probabilities for the first (and only) image
|
68 |
|
69 |
+
# Create a dictionary of label -> score
|
70 |
results = {}
|
71 |
for i, prob in enumerate(probabilities):
|
72 |
label = model.config.id2label[i]
|
|
|
78 |
except Exception as e:
|
79 |
print(f"Error during prediction: {e}")
|
80 |
# Return error in the format expected by gr.Label
|
81 |
+
# Provide a user-friendly error message in the output
|
82 |
+
return {"Error": f"Processing failed. Please try again or use a different image."}
|
83 |
|
84 |
# --- Define Example Images ---
|
85 |
example_dir = "examples"
|
86 |
example_images = []
|
87 |
+
if os.path.exists(example_dir) and os.listdir(example_dir): # Check if dir exists AND is not empty
|
88 |
for img_name in os.listdir(example_dir):
|
89 |
if img_name.lower().endswith(('.png', '.jpg', '.jpeg', '.webp')):
|
90 |
example_images.append(os.path.join(example_dir, img_name))
|
91 |
+
if example_images:
|
92 |
+
print(f"Found examples: {example_images}")
|
93 |
+
else:
|
94 |
+
print("No valid image files found in 'examples' directory.")
|
95 |
else:
|
96 |
print("No 'examples' directory found or it's empty. Examples will not be shown.")
|
97 |
|
98 |
|
99 |
# --- Custom CSS ---
|
|
|
100 |
css = """
|
101 |
body { font-family: 'Inter', sans-serif; } /* Use a clean sans-serif font */
|
102 |
|
|
|
130 |
#prediction-label .label-name { font-weight: bold; font-size: 1.1em; }
|
131 |
#prediction-label .confidence { font-size: 1em; }
|
132 |
|
133 |
+
/* Style the results heading */
|
134 |
+
#results-heading {
|
135 |
+
text-align: center;
|
136 |
+
font-size: 1.2em; /* Slightly larger heading for results */
|
137 |
+
margin-bottom: 10px; /* Space below heading */
|
138 |
+
color: #34495e; /* Match other heading colors */
|
139 |
+
}
|
140 |
+
|
141 |
/* Style the examples section */
|
142 |
.gradio-container .examples-container { padding-top: 15px; }
|
143 |
.gradio-container .examples-header { font-size: 1.1em; font-weight: bold; margin-bottom: 10px; color: #34495e; }
|
|
|
165 |
"""
|
166 |
|
167 |
# --- Gradio Interface using Blocks and Theme ---
|
168 |
+
# Choose a theme: gr.themes.Soft(), gr.themes.Monochrome(), gr.themes.Glass(), etc.
|
|
|
169 |
theme = gr.themes.Soft(
|
170 |
primary_hue="emerald", # Color scheme based on emerald green
|
171 |
secondary_hue="blue",
|
|
|
174 |
spacing_size=gr.themes.sizes.spacing_lg, # More spacing
|
175 |
).set(
|
176 |
# Further fine-tuning
|
177 |
+
body_background_fill="#f8f9fa", # Very light grey background
|
178 |
block_radius="12px",
|
179 |
)
|
180 |
|
|
|
200 |
submit_button = gr.Button("π Classify Image", variant="primary") # Make button prominent
|
201 |
|
202 |
with gr.Column(scale=1, min_width=300, elem_id="output-column"):
|
203 |
+
# Use elem_id and target with CSS for styling
|
204 |
+
gr.Markdown("π **Prediction Results**", elem_id="results-heading")
|
205 |
result_output = gr.Label(
|
206 |
num_top_classes=2,
|
207 |
label="Classification",
|
|
|
209 |
)
|
210 |
|
211 |
# Examples Section
|
212 |
+
if example_images: # Only show examples if they exist and list is not empty
|
213 |
gr.Examples(
|
214 |
examples=example_images,
|
215 |
inputs=image_input,
|
|
|
223 |
gr.Markdown(
|
224 |
"""
|
225 |
---
|
|
|
226 |
This application uses a fine-tuned [SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip) vision model
|
227 |
specifically trained to differentiate between images generated by Artificial Intelligence and those created by humans.
|
228 |
|
229 |
+
You can find the model card here: <a href='https://huggingface.co/{model_id}' target='_blank'>{model_id}</a>
|
|
|
|
|
|
|
230 |
|
231 |
+
Fine tuning code available at [https://exnrt.com/blog/ai/fine-tuning-siglip2/](https://exnrt.com/blog/ai/fine-tuning-siglip2/).
|
232 |
""".format(model_id=MODEL_IDENTIFIER),
|
233 |
elem_id="app-footer"
|
234 |
)
|
235 |
|
236 |
# Connect the button click or image change to the prediction function
|
237 |
+
# Use api_name for potential API usage later
|
238 |
+
submit_button.click(fn=classify_image, inputs=image_input, outputs=result_output, api_name="classify_image_button")
|
239 |
+
image_input.change(fn=classify_image, inputs=image_input, outputs=result_output, api_name="classify_image_change")
|
240 |
|
241 |
|
242 |
# --- Launch the App ---
|
243 |
if __name__ == "__main__":
|
244 |
print("Launching Gradio interface...")
|
245 |
+
iface.launch() # Add share=True for temporary public link if needed: iface.launch(share=True)
|
246 |
print("Gradio interface launched.")
|