Nowadays, in the general trend of online video teaching, students' demand for obtaining valuable information in massive teaching videos has greatly increased. In response to this problem, this study proposes a calculation method for extracting key content areas based on teacher gestures in teaching scenarios. The method includes two steps: first, by the teacher gesture information extracted with the help of human body pose estimation technology we use a certain algorithm to find the key content area (target area), and then we use a two-dimensional Gaussian weighting model method with the image saliency principle to carry out the original image saliency calculation. This method is suitable for most teaching scenarios. It can sort out key content for students in the class, extract valuable information, and can be further applied to teaching quality evaluation and video teaching content recommendation.