How to get the confidence score for pose and hand solution from python script?

[NahianAlindo]

I was wondering how I can get the confidence scores for the boxes and key points for the pose estimation solution and hand tracking solution from a python script. Please help me out as I am a new intern, and having a hard time understanding.

[rmothukuru]

@NahianAlindo,
Can you please elaborate your point. Relevant code or any references will help us understand the issue better. Thanks!

[NahianAlindo]

@NahianAlindo,
Can you please elaborate your point. Relevant code or any references will help us understand the issue better. Thanks!

Actually, my intention is to retrieve the confidence score of the detected keypoints generated by MediaPipe. As for some references, I found out that I needed to look into some messages from the protocol buffers as suggested by #644 It's a closed issue, so I thought I'd post here. My codebase is in python, so I need to get the confidence score using Python. I am new to protobufs. But till now, I have tried the below code snippet by looking at packet_getter.py

`
detection = detection_pb2.Detection()

print(detection.score)
`

But all I get are empty lists.
I am unable to understand what to do next. Please help me out here.

[riadoshi]

I have the same issue -- how can I obtain the tracking confidence of the hands detected? I set the min_tracking_confidence to a value when starting the program, but how do I access the actual value of the tracking confidence?

Thanks for your help!

[jiuqiant]

You can get palm_detection by following #1390 (comment).
The detection messages has a score field: https://github.com/google/mediapipe/blob/ecb5b5f44ab23ea620ef97a479407c699e424aa7/mediapipe/framework/formats/detection.proto#L34