mediaflow-lite gaze weights (InsightFace)

This repository mirrors the InsightFace gaze-estimation assets used by mediaflow-lite (family gaze in mediaflow_lite/models/manifest.yaml).

Files

File Description
latest_a.ckpt PyTorch Lightning checkpoint — Generalizing Gaze Estimation with Weak-Supervision from Synthetic Views (arXiv:2212.02997).
eyes3d.pkl Eye mesh / landmark statistics used with the checkpoint.

Upstream implementation and original Google Drive links: deepinsight/insightface/reconstruction/gaze

Terms of use

InsightFace distributes code under the MIT License, but states that training data and models trained with that data are for non-commercial research purposes only. See the License section of the InsightFace README.

By downloading these files you agree to comply with InsightFace’s terms for pretrained models. This Hub repo is a convenience mirror for reproducible academic research; it is not affiliated with InsightFace. For commercial use, contact the InsightFace authors.

Citation

@article{insightface_gaze_2022,
  title   = {Generalizing Gaze Estimation with Weak-Supervision from Synthetic Views},
  journal = {arXiv preprint arXiv:2212.02997},
  year    = {2022},
}
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for kmu-hsg/mediaflow-lite-gaze