Research map / Video / Video segmentation and tracking
Factorized Higher-Order CNNs With an Application to Spatio-Temporal Emotion Estimation
Jean Kossaifi, Antoine Toisoul, Adrian Bulat, Yannis Panagakis, Timothy M. Hospedales, Maja Pantic
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace regular ResNet-18 convolutions with factorized HO-CPConv blocks | ResNet-18 on AffectNet · AffectNet | valence CCC: 0.66 -> 0.71 | Improved |
| replace regular ResNet-18 convolutions with factorized HO-CPConv blocks | ResNet-18 on AffectNet · AffectNet | arousal CCC: 0.57 -> 0.63 | Improved |
| replace regular convolutions with HO-CPConv blocks | ResNet-18 on SEWA static · SEWA static | valence CCC: 0.29 -> 0.6 | Improved |
| transduce the static HO-CPConv model to video | HO-CPConv on SEWA · SEWA temporal | valence CCC: 0.54 -> 0.75 | Improved |
| transduce the static HO-CPConv model to video | HO-CPConv on AFEW-VA · AFEW-VA temporal | valence CCC: 0.11 -> 0.57 | Improved |
| replace MobileNet-v2 blocks with the proposed 2D factorized blocks | MobileNet-v2 on CIFAR-10 · CIFAR-10 | accuracy: $94$ -> $94$ | No clear effect |
| add a skip connection and use PReLU instead of ReLU (ablation) | HO-CP ConvNet on 20BN-Jester · 20BN-Jester validation | Top-1 accuracy: $83.8$ -> $85.4$ | Improved |
| add a skip connection and use PReLU instead of ReLU (ablation) | HO-CP ConvNet on 20BN-Jester · 20BN-Jester validation | Top-5 accuracy: $97.4$ -> $98.6$ | Improved |
Closest papers by meaning
- MDAN: Multi-Level Dependent Attention Network for Visual Emotion Analysis (CVPR 2022 · similarity 0.87)
- Building a Large Scale Dataset for Image Emotion Recognition: The Fine Print and The Benchmark (AAAI 2016 · similarity 0.87)
- Convolutional Tensor-Train LSTM for Spatio-Temporal Learning (NeurIPS 2020 · similarity 0.87)
- Contemplating Visual Emotions: Understanding and Overcoming Dataset Bias (ECCV 2018 · similarity 0.86)
- MIMAMO Net: Integrating Micro- and Macro-Motion for Video Emotion Recognition (AAAI 2020 · similarity 0.86)
- Deep Rank-One Tensor Functional Factorization for Multi-Dimensional Data Recovery (AAAI 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
