Research & Development Projects
A collection of research and development projects spanning AI, computer vision, game development, and intelligent document processing. All source code is available on GitHub.
Intelligent Document Processing
Python OpenCV OCR
A research testzone dealing with Python OpenCV optical character recognition (OCR). The project combines conventional and deep machine learning techniques, using TensorFlow + Keras, and features a PyQt6 graphical interface. The next phase explores cursive handwriting recognition.
- Technologies: Python, OpenCV, TensorFlow, Keras, PyQt6, Jupyter Notebook
- Focus: Optical Character Recognition (OCR), machine learning
- Status: Active research (49 commits)
PyTorch Handwritten Mathematical Expression Recognition (HMER)
A deep learning research project implementing Handwritten Mathematical Expression Recognition (HMER) using Attention and Coverage mechanisms in PyTorch. Forked and extended with additional improvements including CPU compatibility and file output logging. Tested on the CROHME 2016 dataset, achieving an ExpRate of 38.595%.
- Technologies: Python, PyTorch, DenseNet, Attention RNN
- Dataset: CROHME 2016
- Results: WER loss 17.160%, ExpRate 38.595%
- Status: Active development (111 commits, 8 commits ahead of upstream)
- License: MIT
Current Projects
Kanji Go (漢字GO)
A learning game for Japanese language learners. Best for teenagers and young adults who would like to use their spare time to challenge on the phone.
Upcoming Projects
Music Pom Pom 2