Papers by Mushaffa Rasyid Ridha
Refining rtMRI Landmark-Based Vocal Tract Contour Labels with FCN-Based Smoothing and Point-to-Curve Projection (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods for labeling articulator contours are based on a ground truth label, but there are occasional errors. |
| Approach: | They propose to refine landmark-based vocal-tract contour labels using outlier removal, full convolutional network and a landmark point-to-edge curve projection technique. |
| Outcome: | The proposed labels outperform existing labels and their accuracy through subjective assessments of several contour areas. |