智慧型生理疲勞偵測模組建置研究

The Study of Establishment of the Intelligent Physiological Fatigue Monitoring Module

職場疲勞為近年來常討論之議題，許多行業之工作環境易造成勞工生理疲勞，如駕駛人員、裝配線作業員等，疲勞導致專注力下降，增加工作危害風險。目前影像辨識及分析技術已逐漸普及，本研究將此技術應用於職場疲勞偵測，應用範圍可包含駕駛等易產生作業疲勞之行業勞工。本研究建立之智慧型生理疲勞偵測技術部分，主要功能包含臉部、眼睛、嘴部等特徵建立辨識模組，並建立演算法提供個人疲勞等級。根據測試結果，可提供82% 靈敏度與64%特異性，本監測模組上線使用時，系統可透過串流影像，即時判斷使用者疲勞狀態。在場域模擬與實場測試部份，本研究亦依照石化業（ 含關聯產業） 勞工工作範圍，針對（1） 室內（ 中控室）、（2） 室內其他環境、（3） 室外高空作業環境與（4） 大型運輸車輛工作環境進行系統的場域與實場驗證，驗證結果偵測效果良好。本研究透過智慧型生理疲勞監測系統進行即時疲勞判斷，駕駛作業勞工可同時配戴智慧型手環，或進一步提供健檢資料，生理資訊（影像、手環資料、健檢資料）更可以完整記錄。本系統提供職場疲勞即時量測及異常警示提醒與建議，可改善自身健康狀況，在企業管理部分，本系統未來也可應用於協助降低職業災害發生之機率，且有助於提早發現問題，降低企業財產損失。

Workplace fatigue is a frequently discussed topic in recent years. The working environment ofmany industries is prone to cause physical fatigue of labor, such as drivers, assembly line operators,etc. Fatigue leads to decreased concentration and increases the risk of work hazards. At present, imagerecognition and analysis technology has gradually become popular. This research applies this technologyto workplace fatigue detection. The application scope can include driving and other industries that areprone to work fatigue.In the intelligent physiological fatigue monitoring module part, the main functions include theestablishment of recognition modules for facial, eye, mouth and other features, and the establishment ofalgorithms to provide personal fatigue levels. According to the test results, it can provide 82% sensitivityand 64% characteristics. When this monitoring module is used online, the system can judge the fatiguestatus of users in real time through streaming images. In the field simulation and field test part, thisresearch is also based on the labor scope of the petrochemical industry (including related industries),targeting (1) indoor (central control room), (2) other indoor environments, and (3) outdoor aerial workenvironments. (4) Systematic field and field verification with the working environment of large-scaletransportation vehicles, the detection results of the verification results are good.This study used smart physiological fatigue monitoring system to perform real-time fatiguejudgment. Driving workers can wear smart bracelets at the same time, or provide further healthexamination data, and physiological information (images, bracelet data, health examination data) can becompletely recorded.According to the survey results of the company's acceptance, 92% of workers agree with thisfunction, and also think that abnormal warnings and suggestions can improve themselves. Health status,and according to the survey of corporate acceptance, most management employees think that thissystem can reduce the probability of occupational disasters and help to identify problems early to reducecorporate property losses.