Problems with choosing the duration of a test video clip during the testing phase

[xu-xh]

Hi! Thanks for your work here. Why was the length of the real PPG signal captured during the test 30 seconds instead of a smaller value, 10s or 20s. wouldn't a smaller value mean a closer approximation to real time heart rate. Isn't this relationship the same as the difference between instantaneous rate and average rate?

[zhaodongsun]

Hi, you can choose other lengths for test. 30s is just a default value used in most works. Smaller length is more close to instantaneous heart rate, but may also cause error due to small length.

[xu-xh]

Thanks. And what about the length of the frames and the image size of the model input? In fact, when I set the frame length of the model input to 128 and the image size to 96, your model predictions become much worse, especially on the PURE dataset. Can you please tell me how you compare your proposed contrast-phys+ method between many studies that are in different processing conditions.