This repository contains the easy-to-use pipeline implementation of the FaceXFormer, a unified transformer model for comprehensive facial analysis, as described in the paper by Kartik Narayan et al. from Johns Hopkins University.
Here is official code repo : https://github.com/Kartik-3004/facexformer
You can use FaceXFormer to extract
- landmarks
- headpose orientation
- various attributes
- visibility
- age-gender-race information really fast and from unified model. And you can do it really fast(37 FPS).
pip install facexformer_pipeline To use the FaceXFormer pipeline, follow these steps:
#Import the pipeline class:
from facexformer_pipeline import FacexformerPipeline
#Initialize the pipeline with desired tasks:
pipeline = FacexformerPipeline(debug=True, tasks=['headpose', 'landmark', 'attributes'])
#Run the model on an image:
results = pipeline.run_model(image_array)
#Access the results:
print(results['headpose'])
print(results['landmark_list'])This implementation is based on the research done by Kartik Narayan and his team at Johns Hopkins University. All credit for the conceptual model and its validation belongs to them.