Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -3,7 +3,25 @@ import os
|
|
3 |
from PIL import Image
|
4 |
from transformers import AutoModelForImageClassification, SiglipImageProcessor
|
5 |
import gradio as gr
|
6 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
|
8 |
# Model path
|
9 |
MODEL_PATH = "./model"
|
@@ -12,30 +30,27 @@ try:
|
|
12 |
print(f"=== Loading model from: {MODEL_PATH} ===")
|
13 |
print(f"Available files: {os.listdir(MODEL_PATH)}")
|
14 |
|
15 |
-
# Load the model
|
16 |
print("Loading model...")
|
17 |
model = AutoModelForImageClassification.from_pretrained(MODEL_PATH, local_files_only=True)
|
18 |
print("β
Model loaded successfully!")
|
19 |
|
20 |
-
# Load
|
21 |
print("Loading image processor...")
|
22 |
try:
|
23 |
-
# Try to load the image processor from your local files
|
24 |
processor = SiglipImageProcessor.from_pretrained(MODEL_PATH, local_files_only=True)
|
25 |
print("β
Image processor loaded from local files!")
|
26 |
except Exception as e:
|
27 |
print(f"β οΈ Could not load local processor: {e}")
|
28 |
print("Loading image processor from base SigLIP model...")
|
29 |
-
# Fallback: load processor from base model online
|
30 |
processor = SiglipImageProcessor.from_pretrained("google/siglip-base-patch16-224")
|
31 |
print("β
Image processor loaded from base model!")
|
32 |
|
33 |
-
# Get labels
|
34 |
if hasattr(model.config, 'id2label') and model.config.id2label:
|
35 |
labels = model.config.id2label
|
36 |
print(f"β
Found {len(labels)} labels in model config")
|
37 |
else:
|
38 |
-
# Create generic labels if none exist
|
39 |
num_labels = model.config.num_labels if hasattr(model.config, 'num_labels') else 2
|
40 |
labels = {i: f"class_{i}" for i in range(num_labels)}
|
41 |
print(f"β
Created {len(labels)} generic labels")
|
@@ -44,19 +59,40 @@ try:
|
|
44 |
|
45 |
except Exception as e:
|
46 |
print(f"β Error loading model: {e}")
|
47 |
-
print("\n=== Debug Information ===")
|
48 |
print(f"Files in model directory: {os.listdir(MODEL_PATH)}")
|
49 |
raise
|
50 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
def classify_meme(image: Image.Image):
|
52 |
"""
|
53 |
-
Classify meme and extract text
|
54 |
"""
|
55 |
try:
|
56 |
-
#
|
57 |
-
|
|
|
|
|
|
|
58 |
|
59 |
-
# Process image for
|
60 |
inputs = processor(images=image, return_tensors="pt")
|
61 |
|
62 |
# Run inference
|
@@ -95,13 +131,13 @@ demo = gr.Interface(
|
|
95 |
gr.Label(num_top_classes=5, label="Meme Classification"),
|
96 |
gr.Textbox(label="Extracted Text", lines=3)
|
97 |
],
|
98 |
-
title="π Meme Classifier with
|
99 |
-
description="""
|
100 |
-
Upload a meme image to
|
101 |
-
|
102 |
-
|
103 |
|
104 |
-
Your model
|
105 |
""",
|
106 |
examples=None,
|
107 |
allow_flagging="never"
|
@@ -113,4 +149,4 @@ if __name__ == "__main__":
|
|
113 |
server_name="0.0.0.0",
|
114 |
server_port=7860,
|
115 |
share=False
|
116 |
-
)
|
|
|
3 |
from PIL import Image
|
4 |
from transformers import AutoModelForImageClassification, SiglipImageProcessor
|
5 |
import gradio as gr
|
6 |
+
|
7 |
+
# Alternative OCR using transformers
|
8 |
+
def setup_alternative_ocr():
|
9 |
+
"""Setup alternative OCR using transformers models"""
|
10 |
+
try:
|
11 |
+
from transformers import TrOCRProcessor, VisionEncoderDecoderModel
|
12 |
+
print("Setting up TrOCR for text extraction...")
|
13 |
+
|
14 |
+
ocr_processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-printed")
|
15 |
+
ocr_model = VisionEncoderDecoderModel.from_pretrained("microsoft/trocr-base-printed")
|
16 |
+
|
17 |
+
print("β
TrOCR model loaded successfully!")
|
18 |
+
return ocr_processor, ocr_model, True
|
19 |
+
except Exception as e:
|
20 |
+
print(f"β οΈ Could not load TrOCR: {e}")
|
21 |
+
return None, None, False
|
22 |
+
|
23 |
+
# Try to setup OCR
|
24 |
+
OCR_PROCESSOR, OCR_MODEL, OCR_AVAILABLE = setup_alternative_ocr()
|
25 |
|
26 |
# Model path
|
27 |
MODEL_PATH = "./model"
|
|
|
30 |
print(f"=== Loading model from: {MODEL_PATH} ===")
|
31 |
print(f"Available files: {os.listdir(MODEL_PATH)}")
|
32 |
|
33 |
+
# Load the model
|
34 |
print("Loading model...")
|
35 |
model = AutoModelForImageClassification.from_pretrained(MODEL_PATH, local_files_only=True)
|
36 |
print("β
Model loaded successfully!")
|
37 |
|
38 |
+
# Load image processor
|
39 |
print("Loading image processor...")
|
40 |
try:
|
|
|
41 |
processor = SiglipImageProcessor.from_pretrained(MODEL_PATH, local_files_only=True)
|
42 |
print("β
Image processor loaded from local files!")
|
43 |
except Exception as e:
|
44 |
print(f"β οΈ Could not load local processor: {e}")
|
45 |
print("Loading image processor from base SigLIP model...")
|
|
|
46 |
processor = SiglipImageProcessor.from_pretrained("google/siglip-base-patch16-224")
|
47 |
print("β
Image processor loaded from base model!")
|
48 |
|
49 |
+
# Get labels
|
50 |
if hasattr(model.config, 'id2label') and model.config.id2label:
|
51 |
labels = model.config.id2label
|
52 |
print(f"β
Found {len(labels)} labels in model config")
|
53 |
else:
|
|
|
54 |
num_labels = model.config.num_labels if hasattr(model.config, 'num_labels') else 2
|
55 |
labels = {i: f"class_{i}" for i in range(num_labels)}
|
56 |
print(f"β
Created {len(labels)} generic labels")
|
|
|
59 |
|
60 |
except Exception as e:
|
61 |
print(f"β Error loading model: {e}")
|
|
|
62 |
print(f"Files in model directory: {os.listdir(MODEL_PATH)}")
|
63 |
raise
|
64 |
|
65 |
+
def extract_text_alternative(image):
|
66 |
+
"""Extract text using TrOCR model"""
|
67 |
+
if not OCR_AVAILABLE:
|
68 |
+
return "OCR not available"
|
69 |
+
|
70 |
+
try:
|
71 |
+
# Convert to RGB if needed
|
72 |
+
if image.mode != 'RGB':
|
73 |
+
image = image.convert('RGB')
|
74 |
+
|
75 |
+
# Process with TrOCR
|
76 |
+
pixel_values = OCR_PROCESSOR(image, return_tensors="pt").pixel_values
|
77 |
+
generated_ids = OCR_MODEL.generate(pixel_values)
|
78 |
+
generated_text = OCR_PROCESSOR.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
79 |
+
|
80 |
+
return generated_text
|
81 |
+
except Exception as e:
|
82 |
+
return f"OCR error: {str(e)}"
|
83 |
+
|
84 |
def classify_meme(image: Image.Image):
|
85 |
"""
|
86 |
+
Classify meme and extract text
|
87 |
"""
|
88 |
try:
|
89 |
+
# Extract text using alternative OCR
|
90 |
+
if OCR_AVAILABLE:
|
91 |
+
extracted_text = extract_text_alternative(image)
|
92 |
+
else:
|
93 |
+
extracted_text = "OCR not available in this environment"
|
94 |
|
95 |
+
# Process image for classification
|
96 |
inputs = processor(images=image, return_tensors="pt")
|
97 |
|
98 |
# Run inference
|
|
|
131 |
gr.Label(num_top_classes=5, label="Meme Classification"),
|
132 |
gr.Textbox(label="Extracted Text", lines=3)
|
133 |
],
|
134 |
+
title="π Meme Classifier" + (" with TrOCR" if OCR_AVAILABLE else ""),
|
135 |
+
description=f"""
|
136 |
+
Upload a meme image to **classify** its content using your trained SigLIP2_77 model.
|
137 |
+
|
138 |
+
{'β
**Text extraction** available via TrOCR (Microsoft Transformer OCR)' if OCR_AVAILABLE else 'β οΈ **Text extraction** not available'}
|
139 |
|
140 |
+
Your model will predict the category/sentiment of the uploaded meme.
|
141 |
""",
|
142 |
examples=None,
|
143 |
allow_flagging="never"
|
|
|
149 |
server_name="0.0.0.0",
|
150 |
server_port=7860,
|
151 |
share=False
|
152 |
+
)
|