CODEWITHSUNDEEP
pythonjsonpose-estimationkeypointopenpose
Root
Root

Reputation: 1

Which pose estimation model is used to get keypoints in the JSON format specifically as below, code in Python?

{"version": 1.0, "people": [{"face_keypoints": [], "pose_keypoints": [104.818897637795, 34.4436363636364, 0.915185123682022, 83.9055118110236, 80.0581818181818, 0.723944239318371, 42.5826771653543, 79.8254545454545, 0.618412546813488, 34.7716535433071, 142.661818181818, 0.757904663681984, 31.244094488189, 204.8, 0.52992781996727, 122.456692913386, 81.2218181818182, 0.589333228766918, 125.48031496063, 145.92, 0.702833116054535, 131.527559055118, 192, 0.366538248956203, 58.7086614173228, 193.861818181818, 0.298667620576452, 0, 0, 0, 0, 0, 0, 107.842519685039, 196.421818181818, 0.324830377765466, 0, 0, 0, 0, 0, 0, 93.9842519685039, 28.16, 0.932040095329285, 107.086614173228, 26.9963636363636, 0.939965099096298, 71.3070866141732, 32.5818181818182, 0.885046675801277, 0, 0, 0], "hand_right_keypoints": [], "hand_left_keypoints": []}]}

I have seen few pose estimations related to openpose and others too. But I am not getting specifically these types of output JSON for a 2D image I pass of a single person to the model. Any advices or ideas?

Tried openpose but could not implement it, and didn't see for 2D images either! Trying to implement it in a Python code.

Upvotes: -1

Views: 293

Answers (1)

schro
schro

Reputation: 121

Yeah, I have a dataset in a similar format; you'll likely have to reformat the JSON/ground truth labels to a different format. For example for YOLOv8-pose there's a tutorial on how to format your truth .JSON or .txt files: Ultralytics YOLOv8 Docs Pose Estimation Datasets Overview. So would look up dataset formatting for your use case and then you can probably automate reformatting from there.

Upvotes: 1

Related Questions