Sign inSign up
Robust Facial Landmark Detection via Recurrent Attentive-Refinement Networks thumbnail
slide-image
Pause
Mute
Subtitles not available
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Robust Facial Landmark Detection via Recurrent Attentive-Refinement Networks

Published on Oct 24, 20161733 Views

Junliang Xing

In this work, we introduce a novel Recurrent Attentive-Refinement (RAR) network for facial landmark detection under unconstrained conditions, suffering from challenges like facial occlusions and/or po

ECCV 2016 - Amsterdam

Related categories

Computer Vision

Chapter list

Robust facial Landmark detection via Recurrent Attentive-refinement Networks00:00
Problem Introduction00:19
recurrent Attentive-refinement (RAR) Network01:22
Background02:13
RAR Networks03:05
RAR Networks: Deep Feature Learning03:54
RAR Networks: Robust Shape Initalization04:30
RAR Networks: Attention Module - 106:12
RAR Networks: Attention Module - 206:50
RAR Networks: Attention Module - 307:08
RAR Networks: Attention Module - 407:19
RAR Networks: Refinement Module - 107:27
RAR Networks: Refinement Module - 207:43
RAR Networks: Refinement Module - 307:50
Results on 300W08:01
Results on COFW and AFLW08:47
Comparison Studies08:59
Attention Center Selection Frequencies09:38
Sample Attentive Refinement - 110:06
Sample Attentive Refinement - 210:16
Sample Attentive Refinement - 310:23
Sample Attentive Refinement - 410:25
Sample Attentive Refinement - 510:26
Sample Attentive Refinement - 610:27
Sample Attentive Refinement - 710:29
Sample Attentive Refinement - 810:31
Q&A10:32