Percussion therapy has positive effects on the human body, such as increasing strength, improving range of motion and flexibility. This is due to the vibrations it produces, which when applied to the human body, cause vasodilatory responses. The main contribution of this work is a methodology based on infrared thermography and convolutional neural network to evaluate and classify the range of motion in patients undergoing hamstring percussion therapy. It consists of three steps: (1) data preparation, (2) data augmentation, and (3) convolutional neural network design and validation. In the data preparation step, 50 images acquired in the pre- and post-percussion therapy phases were pooled, and then each image was labeled in one of the two classes: Within and outside range of motion. Data augmentation, including techniques such as flipping and contrast adjustment, was used to expand the dataset. A convolutional neural network model was used to classify the images. When evaluated on a test set, an accuracy of 90.48% was obtained for the classification of 2 classes. These results demonstrate the efficacy and robustness of the approach to assess range of motion in patients undergoing hamstring percussion therapy.