Develop a victim detection model in rescue operations using thermal camera and Jetson computing platform

• DOI: 10.1049/icp.2024.4342
• Publication Date: 2025-01-08T10:18:41Z
• Authors: Van-Khoa Pham; Van-Phuc Nguyen; Thanh-Son Nguyen Huu; Duy-An Nguyen; Thanh-Kieu Tran Thi; Chi-Chia Sun

In emergency situations such as fires or natural disasters, the swift and accurate search detection of victims are crucia l for saving lives minimizing damage. However, in low-light conditions, this becomes particularly challenging due to limited visibility. This research utilizes Yolov8 neural network architecture trained on a custom-created database, combined with thermal cameras Jetson Orin embedded computing platform, develop model human environments. Experimental results show that operates stably, capable distinguishing humans from other objects. Additionally, ability count number people within captured frame enables system meet practical needs rescue operations.