Spaces:
Sleeping
Sleeping
| import re | |
| import streamlit as st | |
| from transformers import AutoModel, AutoTokenizer | |
| import io | |
| from PIL import Image | |
| def load_model(): | |
| tokenizer = AutoTokenizer.from_pretrained('srimanth-d/GOT_CPU', trust_remote_code=True) | |
| model = AutoModel.from_pretrained("srimanth-d/GOT_CPU", trust_remote_code=True, low_cpu_mem_usage=True, use_safetensors=True, pad_token_id=151643) | |
| model.eval() | |
| return model, tokenizer | |
| def handle_error(error_message): | |
| #logging.error(error_message) | |
| st.error(f"An error occurred: {error_message}") | |
| def extract_text(image_bytes, ocr_type): | |
| try: | |
| model, tokenizer = load_model() | |
| image = Image.open(io.BytesIO(image_bytes)) | |
| image.save("temp_image.png", format="PNG") | |
| res = model.chat(tokenizer, "temp_image.png", ocr_type=ocr_type) | |
| return res | |
| except Exception as e: | |
| handle_error(f"Error during OCR extraction: {str(e)}") | |
| return None | |
| def search_keyword(extracted_text, keyword): | |
| keyword = re.escape(keyword) | |
| regex_pattern = rf'\b({keyword})\b' | |
| occurrences = len(re.findall(regex_pattern, extracted_text, flags=re.IGNORECASE)) | |
| highlighted_text = re.sub(regex_pattern, r"<span style='color:red'><b>\1</b></span>", extracted_text, flags=re.IGNORECASE) | |
| return highlighted_text, occurrences | |
| def cache_image_ocr(image_bytes, ocr_type): | |
| return extract_text(image_bytes, ocr_type) | |
| def app(): | |
| st.set_page_config(page_title="OCR Tool", layout="wide", page_icon=":chart_with_upwards_trend:") | |
| st.header("Optical Character Recognition for English and Hindi Texts") | |
| st.write("Upload an image below for OCR:") | |
| if 'extracted_text' not in st.session_state: | |
| st.session_state.extracted_text = None | |
| col1, col2 = st.columns([1, 1]) | |
| with col1: | |
| st.subheader("Upload and OCR Extraction") | |
| uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "png", "jpeg"], accept_multiple_files=False) | |
| # Add OCR type selection dropdown | |
| ocr_type = st.selectbox("Select OCR Type:", ["ocr", "format"]) | |
| if uploaded_file is not None: | |
| st.image(uploaded_file, caption='Uploaded Image', use_column_width=True) | |
| image_bytes = uploaded_file.read() | |
| if st.session_state.extracted_text is None: | |
| with st.spinner("Extracting the text..."): | |
| extracted_text = cache_image_ocr(image_bytes, ocr_type) | |
| if extracted_text: | |
| st.success("Text extraction completed!", icon="🎉") | |
| st.session_state.extracted_text = extracted_text | |
| st.write("Extracted Text:") | |
| st.write(extracted_text) | |
| else: | |
| st.error("Failed to extract text. Please try with a different image.") | |
| else: | |
| st.write("Extracted Text:") | |
| st.write(st.session_state.extracted_text) | |
| else: | |
| st.session_state.extracted_text = None | |
| st.info("Please upload an image file to proceed.") | |
| with col2: | |
| st.subheader("Keyword Search") | |
| if st.session_state.extracted_text: | |
| keyword = st.text_input("Enter keyword to search") | |
| if keyword: | |
| with st.spinner(f"Searching for '{keyword}'..."): | |
| highlighted_text, occurrences = search_keyword(st.session_state.extracted_text, keyword) | |
| if occurrences > 0: | |
| st.success(f"Found {occurrences} occurrences of the keyword '{keyword}'!") | |
| st.markdown(highlighted_text, unsafe_allow_html=True) | |
| else: | |
| st.warning(f"No occurrences of the keyword '{keyword}' were found.") | |
| else: | |
| st.info("Please upload an image and extract text first.") | |
| def main(): | |
| try: | |
| app() | |
| except Exception as main_error: | |
| handle_error(f"Unexpected error in the main function: {str(main_error)}") | |
| if __name__ == "__main__": | |
| main() |