search

Fabian-Sc85 / non-invasive-bp-estimation-using-deep-learning

Assessment of non-invasive blood pressure prediction from PPG and rPPG signals using deep learning
GNU General Public License v3.0
151 stars 36 forks source link

How to get a 875 length input form rPPG signal? #12

Closed wang-tf closed 1 year ago

wang-tf commented 1 year ago

The out put BVP singal from POS was 230 length in my case. So how to convert to model input ? Is there a upsample method was used ? Thanks!

Fabian-Sc85 commented 1 year ago

Hello,

thank you for your interest in our work. Can you give me a little bit more information on what you are trying to do? The dataset published on Zenodo already contains the rPPG signals so there should be no need to use the POS algorithm.

wang-tf commented 1 year ago

Thank you for your work! It's very interesing.

The data on Zenodo usefull. I want to test the retrained model on my personal face video. The question is how to convert video data to model input. Thanks!

The rppg data in rPPG-BP-UKL_rppg_7s.h5 is 875 length (7s * 125). Is that mean the video data mast have 125fps ? There is 30 fps in my video.

Fabian-Sc85 commented 1 year ago

Hi, in case you want to use your own camera data to train the neural network you do not necessarily need 120 fps. You can also upsample your signal to 120 fps using standard interpolation methods e.g. provided by SciPy or NumPy packages. However, it is important that your input signal has a length of 7 seconds.

wang-tf commented 1 year ago

OK, I understand it. I will try it later. Thanks so much.