https://github.com/arielle0222/autonomous-pedestrian-prediction-with-talchum-motion-data
🇰🇷🎭🕺 Leveraging traditional Korean Talchum dance data, this project enhances pedestrian prediction models for autonomous vehicles by capturing complex, non-standard movement patterns.
https://github.com/arielle0222/autonomous-pedestrian-prediction-with-talchum-motion-data
autonomous-driving mask-rcnn object-detection pedestrian-detection python yolov8
Last synced: 7 months ago
JSON representation
🇰🇷🎭🕺 Leveraging traditional Korean Talchum dance data, this project enhances pedestrian prediction models for autonomous vehicles by capturing complex, non-standard movement patterns.
- Host: GitHub
- URL: https://github.com/arielle0222/autonomous-pedestrian-prediction-with-talchum-motion-data
- Owner: Arielle0222
- Created: 2024-11-04T06:13:58.000Z (12 months ago)
- Default Branch: main
- Last Pushed: 2024-12-22T09:34:51.000Z (10 months ago)
- Last Synced: 2025-02-14T14:43:36.277Z (8 months ago)
- Topics: autonomous-driving, mask-rcnn, object-detection, pedestrian-detection, python, yolov8
- Language: Jupyter Notebook
- Homepage:
- Size: 12.3 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Predictive Pedestrian Detection in Autonomous Driving Using Traditional Korean Mask Dance (Talchum) Motion Data 🇰🇷🎭🕺
### 📌 Project Overview
Predicting unexpected pedestrian behavior is essential for safe autonomous driving. This project leverages complex motion patterns from **Talchum (traditional Korean mask dance) data to build a model that can predict non-standard pedestrian behavior**
, which is often missed by conventional pedestrian datasets. By combining YOLO with a keypoint-based detection system, this project aims to enable autonomous vehicles to respond more swiftly and accurately to unpredictable pedestrian actions.
### 🚀 Key Features
- **Complex Motion Learning:** Learns various movement patterns in Talchum, enhancing the model's ability to detect unusual pedestrian behavior.
- **YOLO & Keypoint-based Risk Detection:** Uses YOLO for pedestrian location detection and keypoints for anticipating sudden movements.
### 📊 Data Processing & Analysis
Key and Keypoint Analysis: Visualizes the relationships between height and keypoints using vector flow and heatmaps to understand movement patterns.
Correlation Analysis: Shows correlations between height and keypoints for deeper insight into pedestrian behaviors.
### 📼 Presentation ([watch here!](https://youtu.be/ZgqAdSvPlhI))
### 🛠️ Tools Used
- **Python**: Core programming language for data processing, model training, and analysis.
- **Google Colab**: Cloud environment for training and experimentation.
- **YOLO (You Only Look Once)**: Primary tool for bounding box-based object detection.
- **OpenCV**: Used for computer vision processing and keypoint analysis.
- **Matplotlib & Seaborn**: Libraries for data visualization, heatmaps, and flow mapping.
### 📄 References
- [Hybrid Camera for Self-Driving Car Safety - EETimes](https://www.eetimes.eu/hybrid-camera-targets-self-driving-car-safety/)
- [Hybrid Camera Technology in Self-Driving - Dong-A Science](https://m.dongascience.com/news.php?idx=65680)
- [AI Hub Talchum Dataset - AI Hub](https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=data&dataSetSn=71759)
- [Self-Driving Technology Report - ScienceON](https://scienceon.kisti.re.kr/srch/selectPORSrchReport.do?cn=TRKO202300005284)
- [AIHub Data and Google Colab Usage - Velog](https://velog.io/@guswns7451/%EB%A8%B8%EC%8B%A0%EB%9F%AC%EB%8B%9D-AIHub-%EB%8D%B0%EC%9D%B4%ED%84%B0-%EB%8B%A4%EC%9A%B4-Google-Colab-%EC%82%AC%EC%9A%A9)
- [Predicting Pedestrian Behavior with Hybrid Camera Technology - Yonhap News](https://www.yna.co.kr/view/MYH20211011013700641)