Facial images and landmark coordinate annotations for MobileNet V1-based detection models. These assets provide spatial feature points to support the development of lightweight deep neural network architectures for mobile inference.
Use Cases
- Train a regression model to predict facial keypoints using the landmark coordinate labels.
- Optimize MobileNet V1 weights for real-time facial feature tracking on mobile hardware.
- Evaluate the accuracy of DNN-based landmark detectors against ground-truth facial coordinates.
Strengths
- Includes facial landmark coordinate labels for keypoint regression.
- Specifically formatted for MobileNet V1 architecture compatibility.
- Supports Deep Neural Network (DNN) training for facial feature localization.