Facial Expression Recognition (FER)
Facial Expression Recognition (FER) is a computer vision task aimed at identifying and categorizing emotional expressions depicted on a human face. The goal is to automate the process of determining emotions in real-time, by analyzing the various features of a face such as eyebrows, eyes, mouth, and other features, and mapping them to a set of emotions such as anger, fear, surprise, sadness and happiness.
( Image credit: DeXpression )
Papers
Showing 1–25 of 492 papers
All datasetsAffectNetRAF-DBFER2013FER+Acted Facial Expressions In The Wild (AFEW)CK+FERPlusJAFFESFEWAff-Wild2BP4DDISFA
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | ResEmoteNet | Accuracy (7 emotion) | 72.93 | — | Unverified |
| 2 | Norface | Accuracy (8 emotion) | 68.69 | — | Unverified |
| 3 | EmoAffectNet | Accuracy (7 emotion) | 66.49 | — | Unverified |
| 4 | Emotion-GCN | Accuracy (7 emotion) | 66.46 | — | Unverified |
| 5 | FaceBehaviorNet | Accuracy (7 emotion) | 65.4 | — | Unverified |
| 6 | Ada-DF | Accuracy (7 emotion) | 65.34 | — | Unverified |
| 7 | EAC | Accuracy (7 emotion) | 65.32 | — | Unverified |
| 8 | PAENet | Accuracy (7 emotion) | 65.29 | — | Unverified |
| 9 | DACL | Accuracy (7 emotion) | 65.2 | — | Unverified |
| 10 | DDAMFN++ | Accuracy (8 emotion) | 65.04 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | ResEmoteNet | Overall Accuracy | 94.76 | — | Unverified |
| 2 | FMAE | Overall Accuracy | 93.45 | — | Unverified |
| 3 | QCS | Overall Accuracy | 93.02 | — | Unverified |
| 4 | Norface | Overall Accuracy | 92.97 | — | Unverified |
| 5 | S2D | Overall Accuracy | 92.57 | — | Unverified |
| 6 | BTN | Overall Accuracy | 92.54 | — | Unverified |
| 7 | GReFEL | Overall Accuracy | 92.47 | — | Unverified |
| 8 | DDAMFN++ | Overall Accuracy | 92.34 | — | Unverified |
| 9 | DCJT | Overall Accuracy | 92.24 | — | Unverified |
| 10 | POSTER++ | Overall Accuracy | 92.21 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | EfficientFER | Accuracy | 82.47 | — | Unverified |
| 2 | FERNeXt-SDAFE | Accuracy | 81.33 | — | Unverified |
| 3 | ResEmoteNet | Accuracy | 79.79 | — | Unverified |
| 4 | Ensemble ResMaskingNet with 6 other CNNs | Accuracy | 76.82 | — | Unverified |
| 5 | Mini-ResEmoteNet (A) | Accuracy | 76.33 | — | Unverified |
| 6 | EmoNeXt | Accuracy | 76.12 | — | Unverified |
| 7 | Segmentation VGG-19 | Accuracy | 75.97 | — | Unverified |
| 8 | Local Learning Deep+BOW | Accuracy | 75.42 | — | Unverified |
| 9 | LHC-Net | Accuracy | 74.42 | — | Unverified |
| 10 | Residual Masking Network | Accuracy | 74.14 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | PAtt-Lite | Accuracy | 95.55 | — | Unverified |
| 2 | GReFEL | Accuracy | 93.09 | — | Unverified |
| 3 | QCS | Accuracy | 91.85 | — | Unverified |
| 4 | ResNet18 Dense Architecture | Accuracy | 91.41 | — | Unverified |
| 5 | DDAMFN | Accuracy | 90.74 | — | Unverified |
| 6 | KTN | Accuracy | 90.49 | — | Unverified |
| 7 | Vit-base + MAE | Accuracy | 90.18 | — | Unverified |
| 8 | FER-VT | Accuracy | 90.04 | — | Unverified |
| 9 | EAC | Accuracy | 89.64 | — | Unverified |
| 10 | LResNet50E-IR | Accuracy | 89.26 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | ResNet50 | Accuracy(on validation set) | 65.5 | — | Unverified |
| 2 | LResNet50E-IR (5 models with augmentation) | Accuracy(on validation set) | 65.5 | — | Unverified |
| 3 | EAC | Accuracy(on validation set) | 65.32 | — | Unverified |
| 4 | LResNet50E-IR (1 model with augmentation) | Accuracy(on validation set) | 63.7 | — | Unverified |
| 5 | LResNet50E-IR (1 model) | Accuracy(on validation set) | 61.1 | — | Unverified |
| 6 | Multi-task EfficientNet-B0 | Accuracy(on validation set) | 59.27 | — | Unverified |
| 7 | resnet18_noisy | Accuracy(on validation set) | 55.17 | — | Unverified |
| 8 | resnet18 | Accuracy(on validation set) | 51.18 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | EmoNeXt | Accuracy (8 emotion) | 100 | — | Unverified |
| 2 | PAtt-Lite | Accuracy (7 emotion) | 100 | — | Unverified |
| 3 | ViT + SE | Accuracy (7 emotion) | 99.8 | — | Unverified |
| 4 | FAN | Accuracy (7 emotion) | 99.7 | — | Unverified |
| 5 | Nonlinear eval on SL + SSL puzzling (B0) | Accuracy (7 emotion) | 98.23 | — | Unverified |
| 6 | DeepEmotion | Accuracy (7 emotion) | 98 | — | Unverified |
| 7 | FN2EN | Accuracy (8 emotion) | 96.8 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | KTN | Accuracy(pretrained) | 90.49 | — | Unverified |
| 2 | RAN (VGG-16) | Accuracy(pretrained) | 89.16 | — | Unverified |
| 3 | SENet Teacher | Accuracy(pretrained) | 88.88 | — | Unverified |
| 4 | Local Learning Deep + BOW | Accuracy(pretrained) | 87.76 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | TL | Accuracy | 99.52 | — | Unverified |
| 2 | GReFEL | Accuracy | 96.67 | — | Unverified |
| 3 | ViT | Accuracy | 94.83 | — | Unverified |
| 4 | DeepEmotion | Accuracy | 92.8 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Ada-DF | Accuracy | 60.46 | — | Unverified |
| 2 | RAN (VGG16+ResNet18) | Accuracy | 56.4 | — | Unverified |
| 3 | ViT + SE | Accuracy | 54.29 | — | Unverified |
| 4 | Island Loss | Accuracy | 52.52 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | GReFEL | Accuracy | 72.48 | — | Unverified |
| 2 | EmoAffectNet LSTM | UAR | 52.9 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Norface | ICC | 0.74 | — | Unverified |
| 2 | Ours (VGG-F) | ICC | 0.72 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Norface | ICC | 0.67 | — | Unverified |
| 2 | Ours (VGG-F) | ICC | 0.6 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | DeepEmotion | Accuracy | 99.3 | — | Unverified |
| 2 | GReFEL | Accuracy | 98.18 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | DeXpression | Accuracy | 98.63 | — | Unverified |
| 2 | Facial Motion Prior Network | Accuracy | 82.74 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Dynamic MTL | Accuracy (10-fold) | 89.6 | — | Unverified |
| 2 | PPDN | Accuracy (10-fold) | 84.59 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Covariance Pooling | Accuracy | 87 | — | Unverified |
| 2 | Multi Label Output | Accuracy | 79.26 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Covariance Pooling | Accuracy | 58.14 | — | Unverified |
| 2 | VGG-VD-16 | Accuracy | 54.82 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | EfficientFace | Accuracy | 85.87 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Sequential forward selection | Accuracy | 88.7 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | EmoAffectNet LSTM | UAR | 79 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | ResEmoteNet | Accuracy | 75.67 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | ViT + SE | Accuracy | 87.22 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | EmoAffectNet LSTM | UAR | 69.7 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | EmoAffectNet LSTM | UAR | 82.8 | — | Unverified |