Facial Expression Recognition on CK+ Eight Classes (10-fold cross val)
99.3Avg AccuracyPPDN
Evaluation Results
| Method | Links | |
|---|---|---|
| PPDNExperimental Settings=image-based2016.07 | 99.3 | |
| PPDN w/o peakExperimental Settings=image-based2016.07 | 99.2 | |
| PPDN (standard SGD)Experimental Settings=image-based2016.07 | 99.1 | |
| GoogLeNet (baseline)Experimental Settings=image-based2016.07 | 99 | |
| DTAGN(Joint)Experimental Settings=sequence-based2016.07 | 97.3 | |
| FN2ENTraining Protocol=FaceNet2ExpNet (Fine-Tuned)2016.09 | 96.8 | |
| STM-ExpLetExperimental Settings=sequence-based2016.07 | 94.2 | |
| 3DCNN-DAPExperimental Settings=sequence-based2016.07 | 92.4 | |
| AUDNInput Modality=Image-based2016.09 | 92.1 | |
| VGG-16 face netTraining Protocol=Fine-Tuned (baseline)2016.09 | 89.9 | |
| FN2ENTraining Protocol=Train From Scratch (BN)2016.09 | 88.7 |