병원 낙상 감지 시스템 개발 (Jetson Orin Nano 기반)
Budget: $750 – $1,500 USD
Our team (a university lab) is developing an AI-based fall-detection system for hospital use using a Jetson Orin Nano 8GB dev board. Because falls can be life-threatening for elderly patients, the goal is to accurately recognize only true fall situations — not simple motion or normal bed postures. We plan to base detection on an existing YOLO model, run it in real time on the Jetson board, and trigger immediate alerts when a fall is detected.
Development environment: Ubuntu + Jetson Orin Nano 8GB + IP camera
Target schedule: complete by mid-November 2025
Customize a YOLO-based fall-detection model
Train the model on hospital environment video data to distinguish only “fall” postures.
Minimize confusion with poses such as sitting on or attempting to lie down on a bed.
Produce a high-accuracy inference model suitable for deployment.
Optimize for Jetson Orin Nano
Support Jetson optimizations (CUDA, TensorRT, etc.) to achieve real-time performance.
IP camera integration
Configure video streaming between an RJ45-connected IP camera and the Jetson.
Implement a real-time video preview.
Fall alert system
Provide alerts when a fall is detected via GUI and/or voice/message notifications.
Development environment: Ubuntu + Jetson Orin Nano 8GB + IP camera
Target schedule: complete by mid-November 2025
Customize a YOLO-based fall-detection model
Train the model on hospital environment video data to distinguish only “fall” postures.
Minimize confusion with poses such as sitting on or attempting to lie down on a bed.
Produce a high-accuracy inference model suitable for deployment.
Optimize for Jetson Orin Nano
Support Jetson optimizations (CUDA, TensorRT, etc.) to achieve real-time performance.
IP camera integration
Configure video streaming between an RJ45-connected IP camera and the Jetson.
Implement a real-time video preview.
Fall alert system
Provide alerts when a fall is detected via GUI and/or voice/message notifications.