Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -17,13 +17,10 @@ def load_models():
|
|
| 17 |
RAG, model, processor = load_models()
|
| 18 |
|
| 19 |
# Function for OCR and search
|
|
|
|
| 20 |
def ocr_and_search(image, keyword):
|
| 21 |
# Hardcoded query to extract text in English, Sanskrit, and Hindi
|
| 22 |
text_query = "Extract all the text in English, Sanskrit, and Hindi from the image."
|
| 23 |
-
results = RAG.search(text_query, k=1)
|
| 24 |
-
|
| 25 |
-
if not results: # Check if results are empty
|
| 26 |
-
return "No results found for the given query.", [], "{}"
|
| 27 |
|
| 28 |
# Prepare message for Qwen model
|
| 29 |
messages = [
|
|
@@ -66,6 +63,7 @@ def ocr_and_search(image, keyword):
|
|
| 66 |
|
| 67 |
return extracted_text, matched_sentences, json_output
|
| 68 |
|
|
|
|
| 69 |
# Gradio App function
|
| 70 |
def app(image, keyword):
|
| 71 |
# Call OCR and search function
|
|
|
|
| 17 |
RAG, model, processor = load_models()
|
| 18 |
|
| 19 |
# Function for OCR and search
|
| 20 |
+
# Skip RAG search and use Qwen2VL for direct OCR
|
| 21 |
def ocr_and_search(image, keyword):
|
| 22 |
# Hardcoded query to extract text in English, Sanskrit, and Hindi
|
| 23 |
text_query = "Extract all the text in English, Sanskrit, and Hindi from the image."
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
|
| 25 |
# Prepare message for Qwen model
|
| 26 |
messages = [
|
|
|
|
| 63 |
|
| 64 |
return extracted_text, matched_sentences, json_output
|
| 65 |
|
| 66 |
+
|
| 67 |
# Gradio App function
|
| 68 |
def app(image, keyword):
|
| 69 |
# Call OCR and search function
|