Improving MediaPipe Lite Pose Detection in Fog and Low Light

[PareekHeymun]

Have I written custom code (as opposed to using a stock example script provided in MediaPipe)

None

OS Platform and Distribution

Windows 10 Home

Python Version

3.8.0

MediaPipe Model Maker version

No response

Task name (e.g. Image classification, Gesture recognition etc.)

Pose Detection

Describe the actual behavior

Unable to detect the person in fog

Describe the expected behaviour

To be able to detect the person in fog

Standalone code/steps you may have used to try to get what you need

I donwloaded a video from youtube - "The Grand Tour" where the guy is trying to look for his car in the fog, however even tho I can see the dude, mediapipe doesn't detect it. Only when the dude comes closer, it is able to detect it and create its pose landmarks.

Other info / Complete Logs

Additionally, there is a difference in behaviour observed when I resize the image using cv2
img = cv2.resize(img, (640, 480))
actually created a different BEHAVIOUR than
img = cv2.resize(img, (1300, 1000))
because the latter caused a car to be detected as a human pose!