Code. Learn. Break. Fix. Repeat.


# 👋 Introduction

Hi, I'm Ke He (Koen), an undergraduate student at Wuhan University (WHU).

I stand at the intersection of Computer Vision (CV) and System Engineering. My philosophy is simple: bridge the gap between cutting-edge algorithms and real-world applications. I strive not just to reproduce state-of-the-art methods, but to build systems that are Reproducible, Maintainable, and Deliverable.

  • 🔭 Research Interest: Cross-Modality Person Re-Identification (VI-ReID) & Multi-Agent Systems.
  • 🌱 Goal: To become a pioneer who drives change through AI.

# 🛠 Tech Stack

  • Core: Python · PyTorch · C++
  • Web: TypeScript · React · Node.js
  • Infra: Linux · Docker · HPC / Cluster Environment

# 🚀 Selected Projects

# 🔬 Research & AI

1. Medical Diagnosis System (Knowledge Graph & RAG)

Full-stack medical auxiliary system combining structured knowledge and semantic reasoning.

  • Core: Combines Disease-Symptom Knowledge Graphs (KG) with Vector-based RAG for high-accuracy reasoning.
  • Stack: KnowledgeGraph, NLP, Vector DB.
  • Link: Live Demo

2. YaoZhiYunAn: Multi-Agent Disaster Prediction

Real-time situational awareness for emergency response.

  • Core: Built on a Multi-Agent System architecture for collaborative sensing, information fusion, and interpretable visualization.
  • Stack: Multi-Agent, Public Safety.

3. SamLora: Extreme Environment Housing Detection

Robust detection in complex terrains.

  • Core: Fine-tuned the Segment Anything Model (SAM) and integrated with ArcGIS Pro to adapt to challenging geographical data.
  • Stack: GIS, CV, Fine-Tuning.

4. CV Image Dehazing System

Engineering implementation for low-visibility restoration.

  • Core: A complete pipeline for restoring visibility in hazy images using advanced CV algorithms.
  • Status: Software Copyright (CSC No.2)

5. Multi-Target Trajectory Analysis Platform

Complex scene understanding.

  • Core: Analyzes and correlates multi-target movements for surveillance and tracking.
  • Status: Software Copyright (CSC No.1)

# 💻 Engineering & Products

1. MarkTex: Scientific Markdown Editor

Seamless academic writing experience.

  • Feature: Real-time KaTeX rendering combined with high-fidelity PDF Export. Designed for researchers who need speed and precision.
  • Stack: React, KaTeX.
  • Link: Live Demo

2. GPTManager: Multi-Agent Battle System

AI that thinks deeper through debate.

  • Feature: An innovative "Multi-Agent Battle Mode" where multiple GPT agents debate a topic to synthesize the optimal solution.
  • Stack: LLM, Prompt Engineering.
  • Link: Live Demo

3. POSTOFFER: AI Mock Interview Assistant

Your personal AI career coach.

  • Feature: Generates dynamic interview scenarios and provides structured feedback to help users improve expression and logic.
  • Stack: LLM, Full-Stack.
  • Link: Live Demo

4. AI4Learning

Unified AI Education Platform.

  • Feature: Integrates various AI tools into academic disciplines to empower education efficiency.

5. WenJx

Automation for efficiency.

  • Feature: A Python-based automation tool designed to streamline and auto-fill complex questionnaires.

# 📜 Intellectual Property

  1. Multi-Target Trajectory Correlation Analysis Platform (Software Copyright)
  2. Computer Vision Image Dehazing System (Software Copyright)

# 📫 Connect

更新于

请我喝[茶]~( ̄▽ ̄)~*

koen 微信支付

微信支付

koen 支付宝

支付宝