Constrained Local Neural Fields for Robust Facial Landmark Detection in the Wild

Baltrušaitis, Tadas; Robinson, Peter; Morency, Louis–Philippe

doi:10.1109/iccvw.2013.54

Cited by 348 publications

(248 citation statements)

References 14 publications

Supporting

Mentioning

248

Contrasting

Order By: Relevance

“…Our baseline of face recognition uses the ALEX-AF representation with cosine similarity and softmax score fusion, and the dataset used is CS2. Four state-of-the-art facial landmarks are used -(1) DLIB [9] with 68 points, (2) FPS3K [4] with 68 points, (3) TD-CNN [23] with 5 points, and (4) CLNF with 68 points [2,3] and its variant CLNFs that use seed landmarks provided in CS2 [11] for the estimation of an initial face shape. Note that we use DLIB, FPS3K and TDCNN out-of-shelf, and because they do not provide an interface to set the initial facial landmarks, we cannot report their performance with seed landmarks.…”

Section: Effect Of Landmark Detection Algorithmmentioning

confidence: 99%

Face recognition using deep multi-pose representations

AbdAlmageed

Rawls

et al. 2016

2016 IEEE Winter Conference on Applications of Computer Vision (WACV)

136

View full text Add to dashboard Cite

We introduce our method and system for face recognition using multiple pose-aware deep learning models. In our representation, a face image is processed by several posespecific deep convolutional neural network (CNN) models to generate multiple pose-specific features. 3D rendering is used to generate multiple face poses from the input image. Sensitivity of the recognition system to pose variations is reduced since we use an ensemble of pose-specific CNN features. The paper presents extensive experimental results on the effect of landmark detection, CNN layer selection and pose model selection on the performance of the recognition pipeline. Our novel representation achieves better results than the state-of-the-art on IARPA's CS2 and NIST's IJB-A in both verification and identification (i.e. search) tasks.

show abstract

Section: Effect Of Landmark Detection Algorithmmentioning

confidence: 99%

Face recognition using deep multi-pose representations

AbdAlmageed

Rawls

et al. 2016

2016 IEEE Winter Conference on Applications of Computer Vision (WACV)

136

View full text Add to dashboard Cite

show abstract

“…This method has been improved with the application of facial landmark identification and tracking through the use of a more robust face tracker called the Cambridge Face Tracker [9]. The output of this face tracker, depicted in Figure 1a provides an outline of the face together with a general direction of looking.…”

Section: Face Recognition Modelmentioning

confidence: 99%

iCub Visual Memory Inspector: Visualising the iCub’s Thoughts

Camilleri

Damianou

Jackson

et al. 2016

Biomimetic and Biohybrid Systems

View full text Add to dashboard Cite

show abstract

“…Therefore, the procedure is repeated by changing the direction of comparison to find a matching time step t * 1 in V 1 for each time step t 2 of V 2 . Two sets of matching sub-sequences are obtained as {t 1 m}. Resulting pairs of "2m+1"-frame image sequences are used as inputs in the proposed method.…”

Section: Expression Matchingmentioning

confidence: 99%

“…To this end, a state-of-the-art tracker [2] is used. The tracker uses an extended version of Conditional Local Neural Fields (CLNF) [1], where individual point distribution and patch expert models are learned for eyes, lips and eyebrows. Detected points by individual models are then fit to a joint point distribution model.…”

Section: Facial Landmark Tracking and Alignmentmentioning

confidence: 99%