app demo
app demo
app demo

Webcam Stress Meter

Submitted to MetroHacks2022 under 18

Inspiration

While surfing around google and youtube I found this interesting article about Photoplethysmogram. It tells us that traditional oxy/heart rate monitor use cameras to measure changes in light absorption (corresponding to an absorption peak by (oxy-) hemoglobin) that can be converted to a heart rate. So I think why not use a webcam? While it looks very impossible but it's possible by setting the webcam for high iso sensitivity and capturing a color fluctuation on our skin.

And secondly, there is an article where it states that our heart rate is related to the stress we experience Effects of stress on heart rate complexity—A comparison between short-term and chronic stress. So after a bit of brainstorming, I decided to build a Webcam Stress Meter.

What it does

It calculates your heart rate only using your webcam and combines it using the face emotions model resulting in a calculation for the stress meter. You can use it while you working/studying so that when your stress meter is reaching a high level, the app will alert you and recommend you to take a rest.

How we built it

To build the webcam heart rate monitor I use cv2 to capture, and distribute webcam data, some filters, and to track our faces. Then a numpy gives some calculation to the tensor so that the app can measure the light change in our skin and extract it into a heart rate.

For the face emotions model, I use this kaggle data and train it using google colab with TensorFlow (CNN). then import it to the script using TensorFlow again and with help of cv2, the app can predict a real-time webcam stream.

Then I combine this two to calculate the stress level of the users using basic math & numpy.

Challenges we ran into

it's challenging to integrate the heart rate app and face emotions app, especially when I have to make a function to determine the stress level, but it is all done now, and it's working :))