Remote Photo-plethysmography (rPPG) is a method to obtain blood volume changes based on the principle of diffuse reflection of natural light on human skin. It can obtain physiological information such as human photoplethysmography signal and heart rate non-invasively and without contact. This research proposes a non-contact heart rate estimation method based on U-Net global facial segmentation to solve the problem that existing rPPG detection methods are easily affected by ambient light and individual movement, and fixed facial areas of interest can easily lead to inaccurate heart rate estimation. This method extracts full facial skin from the video through the U-Net segmentation network, then performs skin pixel reorganization, and extracts the green channel information of the reorganized skin to estimate individual heart rate. To verify the effectiveness and accuracy of this method, testing, and ablation experiments were conducted on the UBFC dataset. The average absolute error, root mean square error, and Pearson correlation coefficient were 4.52 bpm, 7.04 bpm, and 0.93 respectively.