SeetaPsych Face Enhanced

Enhanced face modules for SeetaPsych

Usage

This project is already included in the seetapsych-lib default configuration. Download and use it via seetapsych-manager download.

For usage, refer to SeetaPsych.

You can additionally add this algorithm module using the following methods.

WebUI

Run seetapsych-webui with the --dirs argument to use it.

seetapsych-webui --files seetapsych_face_ex/modules/dense_landmarks.yml

Programmatic Usage

Add the following code in your program to use this algorithm module.

from seetapsych_lib.runtime.factory import Factory
from seetapsych_lib.runtime.pipeline import Pipeline

factory = Factory()
factory.load_file_modules('seetapsych_face_ex/modules/dense_landmarks.yml')

pipeline = Pipeline(factory, ...)

pipeline.add_attributes('face/dense_landmarks')

Introduction

SeetaDenseLandmarks

ONNX-based dense facial landmark detection model by SeetaPsych, capable of detecting 280 2D landmarks. Accepts either a face bounding box or sparse landmarks (2/5/280 points) as input. When given a bbox, it first predicts an initial coarse set, then optionally runs a refinement pass aligned via 5-point similarity transform. Network input size is fixed at 192x192.

Module config: dense_landmarks.yml. Provide Attributes: face/dense_landmarks.

Requires: face/detection (primary). Optionally leverages face/landmarks if available for better alignment.

Available model: seeta-face-ex-dense-landmarks-280.onnx (recommended).

Parameters:

• refine (bool, default true): Enable the second-pass refinement using 5-point similarity alignment.