Our prediction process involved two key steps:

1. Image-to-Text Conversion: Extracting textual data from the provided test images using Optical Character Recognition (OCR) technology.
2. Information Extraction with NER: Cleaning and processing the extracted text using Named Entity Recognition (NER) techniques to identify and extract specific information, ensuring a more accurate and targeted prediction.

1. Import Libraries: The code begins by importing necessary libraries such as requests (for HTTP requests), PIL (for handling images), pytesseract (for OCR), and pandas (for data manipulation).
2. Load DataFrame: The program loads a CSV file (test.csv) into a pandas DataFrame, which contains image URLs and relevant information.
3. extract_text_from_image Function:
   • Download Image: Sends a GET request to the given URL to download the image.
   • Open Image: Opens the downloaded image using PIL.Image.open.
   • OCR: Uses pytesseract.image_to_string to extract text from the image.
   • Error Handling: Includes a try-except block to catch errors and return error messages.
4. Apply OCR: The function is applied to the 'image_link' column of the DataFrame, storing the extracted text in a new column called 'tesseract'.
5. Save DataFrame: The updated DataFrame with the extracted text is saved as a new CSV file (csv/101000.csv).

1. Load and Prepare Data: Loads a new CSV file (likely the one generated in Part 1) and defines lists of different unit types (e.g., length, weight, volume, power).
2. standardize_unit Function: Standardizes units to ensure consistency (e.g., "cm" to "centimeter").
3. process_text Function:
   • Lowercase Text: Converts text to lowercase.
   • Entity-Specific Logic: Extracts values based on the 'entity_name' column (e.g., 'depth', 'width', 'item_weight').
   • Regular Expressions: Finds numerical values followed by specific units.
   • Unit Standardization: Calls the standardize_unit function for consistent unit representation.
   • Print Values: Prints the extracted values for debugging purposes.
4. Apply Processing: Applies process_text to each row of the DataFrame, storing results in the 'extracted_value' column.
5. Entity-Unit Mapping: Creates a dictionary entity_unit_map to store unique units found for each entity type.
6. Save Results: Saves the processed DataFrame to a new CSV file (extracted_results_from_combined.csv).

We developed a two-step solution for extracting and standardizing information from product images:

1. Optical Character Recognition (OCR): Using the Pytesseract library, we converted image data into machine-readable text.
   • Downloaded images via URLs.
   • Processed the images to extract textual content.
2. Named Entity Recognition (NER): A rule-based NER system cleaned and analyzed the extracted text.
   • Text Preprocessing: Cleaned and prepared the text.
   • Entity-Specific Rules: Applied logic and regular expressions to extract numerical values and associated units.
   • Unit Standardization: Standardized units (e.g., "cm" to "centimeter") for consistency.

This automated process significantly enhanced the efficiency and accuracy of extracting and standardizing product information from images.