HRNet-Facial-Landmark-Detection
HRNet-Facial-Landmark-Detection copied to clipboard
Model Predictions i.e points are overlapping.
I have trained on custom dataset of type WFLW but most of the predicted points are overlapping, is anyone facing the same issue? can anyone help? I suspect the problem is due to point normalization at line 59. Can anyone suggest what should I do for 4 points.