Document processing involves automating the extraction of structured data from various documents like invoices and resumes, with the main challenge being the accurate and automatic labeling of extracted text. Industries such as finance and e-commerce benefit significantly from this process as it enhances efficiency by saving employee time spent on manual data entry. Advanced models like LayoutLM, which combine text, image, and locational data, are pivotal for this task. LayoutLM employs a unique method by integrating embeddings for text, location, and images, enabling precise data extraction and classification. The model's open-source availability via platforms like Hugging Face facilitates its use for document classification and text labeling tasks. Its successor, LayoutLMv2, introduces improved training objectives and embeddings, achieving better performance by incorporating 1-D spatial and visual token embeddings to align text and image data more effectively.