The purpose of this research is to propose a system that allows users to objectively check the quantified load on their muscles in the exercise training, especially in rehabilitation and strength training. This is because it is difficult to accurately recognize the quantified load on the muscle. And there is a problem that an appropriate effect cannot be obtained from exercise training due to excessive load applied on muscles when exercise training. The proposed method is to estimate the load on muscles from the device using the IR sensor array and gauge sensor, using SVR (Support Vecotr Regression) which is a machine learning algorithm of regression. This is because the IR sensor device used in this research is a wearable device that is not bulky and is robust against noise such as sweat electromagnetic noise. As a conclusion, it can be said that the system we suggested is not bulky. In addition, it can be said that by using a sensor using infrared light, it is possible to estimate the muscle load using a sensor that is robust against noise. And it is found that there is a correlation between muscle bulge and the load on the hands and arms. Since the MAE is 2.88N and the standard deviation is 1.13N, the force applied to the user’s muscles is confirmed during rehabilitation and training using exercise with isometric contraction that does not produce more force than his own power It can be said that this is an error that does not bother user.