← Back to projects
Research & PaperResearch · Computer Vision · OCR · Mobile Applications· 2023 – 2024
OCR Research: Mobile Invoice Parsing
Investigating Tabular Extraction & Document Layout Challenges in Mobile Apps
Status details: Authored academic research paper on mobile OCR extraction
Authored a computer vision research paper examining the intricate obstacles of camera-based OCR invoice recognition on mobile devices. Explored why complex tabular alignments and variable mobile capture conditions break standard OCR pipelines and analyzed multi-stage structural parsing.
Technologies & Stack
PythonOpenCVTensorFlowTesseract OCRYOLO
Architecture & Engineering Highlights
- Detailed analysis of tabular invoice parsing limitations in mobile environments using Tesseract and YOLO.
- Separated character-level OCR optical recognition from logical table structure and key-value pair reconstruction.
- Proposed adaptive image binarization, perspective correction, and layout-aware post-processing pipelines.