Projekt

Stud. Projektgruppe - Data Science Semesterprojekt

Automated Lieferschein Recognition

  • Bildschirmablauf zur digitalen Bearbeitung von Lieferscheinen: Sicheres Anmelden, Hochladen eines PDF-Dokuments, Anzeige der extrahierten Daten in bearbeitbarer Form, Übersichtsliste aller Dokumente mit Filteroptionen und Download der wichtigsten Informationen als kompakte Zusammenfassung. © stud. Projekt

Project Goal 
The goal of this project is to automate the processing of delivery notes from a paper-based archive. A desktop application automatically extracts the relevant informatlon, stores the document, and allows employees to verify or correct the extracted fields. 

Our Solution 
We developed a Windows desktop application that automatically processes delivery notes using a fine-tuned document understanding model. The application extracts the relevant fields, displays the results for verification, and stores the document and data in a database for easy search and retrieval.

  1. PDF Upload: A delivery note (PDF) is uploaded to the application. 
  2. Al Extraction (DONUT): A fine-tuned DONUT model automatically extracts the relevant information from the document. 
  3. Extracted Data: The application extracts eight relevant fields. 
  4. Human Verification: Employees review the extracted information and can correct the fields if needed. 
  5. Database & Search: All documents and extracted data are stored in a database and can be searched, filtered and retrieved easily. 

User Authentication 
A secure login ensures that only authorized employees can access the application. 

Document Processing 
Uploaded PDF files are preprocessed and displayed directly in the application for easy review. 

Field Extraction & Verification
The application extracts the predefined fields, which can be checked and corrected by the employee. 

Management & Archive 
All delivery notes and extracted data are stored in a database. Employees can search, filter and retrieve documents efficially. 

Betreuer/in