• Aucun résultat trouvé

A Novel Space-Time Representation on the Positive Semidefinite Cone for Facial Expression Recognition

N/A
N/A
Protected

Academic year: 2021

Partager "A Novel Space-Time Representation on the Positive Semidefinite Cone for Facial Expression Recognition"

Copied!
11
0
0

Texte intégral

Loading

Figure

Figure 1. Overview of the proposed approach – After automatic landmark detection for each frame of the video, the Gram matrices are computed to build trajectories on S + (2, n)
Figure 2. A pictorial representation of the positive semidefinite cone S + (2, n). Viewing matrices G 1 and G 2 as ellipsoids in R n , the closeness consists of two contributions: d 2 G (squared  Grass-mann distance) and d 2 P 2 (squared Riemannian distanc
Table 1. Confusion matrix of the proposed trajectory representa- representa-tion and classificarepresenta-tion on S + (2, n) – CK+ database.
Figure 3. Accuracy of the proposed approach when varying the weight parameter k, on (left to right) CK+, MMI, Oulu-CASIA, and AFEW.
+2

Références

Documents relatifs

Meng, “Automatic 3d facial expression recognition using geometric and textured feature fusion,” in IEEE International Conference on Automatic Face and Gesture Recognition and

[3] presented a dynamic curvature based approach for facial activity analysis, then constructed the dynamic curvature descriptor from local regions as well as temporal domains, and

Unlike the latter, the barycentric representation allows us to work directly on Euclidean space and apply a metric learning algorithm to find a suitable metric that is

Facial expression recognition, Low Resolution Images, Local Binary Pattern, Image pyramid, Salient facial

Experimental results on the JAFFE acted facial expression database and on the DynEmo spontaneous expression database demonstrate that our algorithm outperforms many recently

Differently from existing approaches, we exploit the local characteristics of the face by computing SIFT descriptors around a small set of facial landmarks identified on range

This study compared the recognition of facial expression with 6 basic expressions on high fWHR faces or on low fWHR faces under different duration conditions (67ms, 333ms, 600ms),

The video analysis performances can be increase by the addition of audio information, typically those presented in this document : Smile , Eyebrows , Head shake