Fatigue and stamina prediction of athletic person on track using thermal facial biomarkers and optimized machine learning algorithm.

Athletic person's fatigue and stamina prediction plays a vital role for improving the overall performance in the sports. Identification of the athletic person's facial expression on track and field using image, is still a challenge task. The complex background and improper environmental lighting conditions affects the identification of athlete's facial expressions while playing. Existing methods use RGB and traditional night vision cameras for detecting athlete's facial expressions that operates only in minimum lighting condition. These cameras does not function in low lighting (< 30%) and complete dark environment. Moreover, the existing systems never predict fatigue, pain and stamina of the player on the ground in dark environment. In this paper, the facial thermal images of athletic person during playing are acquired and enhanced through the proposed HEOP preprocessing method. Further, the proposed ECOC-MCSVM method classifies fatigue, pain and stamina of sportsperson using facial biomarkers such as cheek raising, lip spreading, tongue position, jaw dropping and nose wrinkling. The prediction levels are optimized using Bayesian optimized Multiple Polynomial Regression analysis (BO-MPR). The proposed ECOC-MCSVM method has an accuracy of 97.69% for fatigue, pain and stamina prediction and it is validated with existing methodologies.