This repository collects my reading notes for computer science research papers. The notes are written in Markdown and organized for review in Obsidian or directly on GitHub.
| Folder | Description | Notes |
|---|---|---|
| AI4H | Papers and notes on artificial intelligence for healthcare, clinical analysis, and medical decision support. | Focused AI-for-health reading notes. |
| Basics | Foundational papers and core concepts in AI, ML, and computer vision. | Start here for classic papers. |
| Templates | Reusable templates for future literature notes, organized by module. | Use these to keep note format consistent. |
The note tables use short display names so the README stays readable on GitHub. Full original paper titles, authors, and DOI links are preserved inside each note.
| Note | Year | Venue / Source | Status | Focus |
|---|---|---|---|---|
| Clinical Gait Analysis | 2024 | PLOS Digital Health | Read | Video-based pose estimation for clinical gait analysis. |
| Monocular 3D Gait Assessment | 2021 | Scientific Reports | Read | Monocular smartphone-video gait parameters validated against GAITRite. |
| Spatiotemporal Gait Parameters | 2020 | JMIR mHealth and uHealth | Read | Markerless gait parameters compared with OptoGait and high-speed video. |
| VisionMD-Gait | 2026 | Scientific Reports | Read | Smartphone-video clinical gait assessment. |
| MotionMetrix-Qualisys Agreement | 2023 | Sensors | Read | MotionMetrix markerless gait measurements compared with Qualisys 3D motion capture. |
| Multiple Pose Trackers for Older Adult Gait | 2021 | Journal of NeuroEngineering and Rehabilitation | Read | RGB-video pose tracking for older-adult gait across trackers, camera heights, and walking directions. |
| Smartphone Location for Markerless Gait Analysis | 2024 | Bioengineering | Read | Smartphone camera location effects on markerless pose-estimation gait analysis. |
| Note | Year | Venue / Source | Status | Focus |
|---|---|---|---|---|
| AlexNet | 2012 | NeurIPS | Read | Large-scale CNN image classification on ImageNet. |
The current templates are organized by module:
My paper-reading workflow follows Mu Li's three-pass paper reading method. I first skim the paper to understand the title, abstract, figures, and overall contribution. Then I read more carefully to understand the method, assumptions, experiments, and results. Finally, I revisit the paper in depth when I need to connect it to other work or write a more complete note.
I use Zotero for reading and annotation. Highlight colors have specific meanings:
- Yellow: ordinary important points, definitions, methods, or results.
- Green: parts I do not fully understand yet.
- Red: possible problems, weaknesses, limitations, or questionable claims.
- Blue: strengths, elegant ideas, or things the paper does especially well.
After reading, I use these highlights to fill in the note template, turning annotations into structured summaries, critiques, and follow-up questions.
- Browse notes by folder from the Repository Structure section.
- Open individual paper notes from the Paper Notes table.
- Create new notes from the module-specific templates in Templates.