Project Profile

YOLO Right-of-Way Recognition

結合物件偵測與追蹤，把交通影片中的不禮讓行人情境轉成可檢查事件的大學畢業專題。

Completed
Status
Completed
Portfolio
Featured
Role
Undergraduate capstone · desktop computer-vision tool
GitHub Read overview
Computer VisionObject DetectionOpenCV

YOLO Right-of-Way Recognition 是我的大學畢業專題。它把單格影像中的物件辨識、跨畫面的追蹤與場景規則接在一起，讓交通影片中的「不禮讓行人」不只是一個模糊描述，而是可以回看與檢查的事件標記。

專案範圍

  • 用 YOLOv8 找出行人、車輛、機車與斑馬線
  • 用 DeepSORT 保留物件在連續畫面中的身份
  • 以 OpenCV 與 PyQt5 支援影片處理、旋轉校正、預覽與進度

這個 Profile 聚焦在工具的角色與可檢查輸出；介面截圖與完整流程放在相關 Overview Article。

What changed or was learned

可驗證成果

Turns video into inspectable events

YOLOv8 detections and DeepSORT tracks are combined with scene rules to mark possible right-of-way violations across a video, not just a single frame.

A complete desktop workflow

The PyQt5 interface covers video selection, model selection, rotation correction, preview, and per-file or batch progress.

Traceable sources

Evidence Trail

  1. Overview Article

    Annotated interface captures, rotation correction, and day/night output examples.

    Inspect source

  2. Source repository

    The YOLOv8, DeepSORT, OpenCV, and PyQt5 implementation.

    Inspect source

Writing