● Conducted an extensive research project on the digitalization of forms using machine learning techniques.
● Examined the fields of document AI and form element detection to extract information from different
categories in fillable form such as check boxes, text areas and .
● Built a form element detection model using YOLOv3 architecture, achieving an 84% F1 score in detecting text
boxes, lines, and inputs. An LSTM model was implemented for annotation classification and information
extraction from receipts, obtaining an F1 score of 76% with bounding box information.
● Implemented a LayoutLM architecture to improve comprehension of text and layout information, obtaining a
remarkable F1 score of 95.69 percent on the SROIE 2019 Receipt dataset.