import imageio
import fastface as ff
# checkout available pretrained models
print(ff.list_pretrained_models())
# ["lffd_slim", "lffd_original"]
# build pl.LightningModule using pretrained weights
model = ff.FaceDetector.from_pretrained("lffd_slim")
# set model to eval mode
model.eval()
# load image
img = imageio.imread("<your_img_file_path>")[:, :, :3]
# find faces
(preds,) = model.predict(img)
"""preds
{
'boxes': [[xmin, ymin, xmax, ymax], ...],
'scores':[<float>, ...]
}
"""
# visualize predictions
pil_img = ff.utils.vis.render_predictions(img, preds)
pil_img.show()