search

google-ai-edge / mediapipe

Cross-platform, customizable ML solutions for live and streaming media.
https://mediapipe.dev
Apache License 2.0
26.78k stars 5.09k forks source link

How can I run the face landmark model with attention on lips and eyes to provide more accuracy, and additionally output iris landmarks #5582

Open jhlibra opened 3 weeks ago

jhlibra commented 3 weeks ago

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

None

OS Platform and Distribution

windows

MediaPipe version

10.00

Bazel version

6.41

Solution

Holistic

Programming Language and version

C++

Describe the actual behavior

Whether to run the face landmark model with attention on lips and eyes to provide more accuracy, and additionally output iris landmarks. If unspecified, functions as set to false. (bool),How can I set it to true

Describe the expected behaviour

I want to run the face landmark model with attention on lips and eyes to provide more accuracy, and additionally output iris landmarks

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

# Whether to run the face landmark model with attention on lips and eyes to
# provide more accuracy, and additionally output iris landmarks. If unspecified,
# functions as set to false. (bool)
input_side_packet: "REFINE_FACE_LANDMARKS:refine_face_landmarks"

Other info / Complete Logs

No response

kuaashish commented 2 weeks ago

Hi @jhlibra,

We believe it is currently not possible using our new Task API. Previously, the model with attention was provided in our legacy solution, which is now being upgraded to the new Task API. Please allow us some time, and we will get back to you with an update.

Thank you!!