# Convolutional neural networks pretrained on large face recognition datasets for emotion classification from video

arXiv: https://arxiv.org/abs/1711.04598 · HTML: https://mlsanity.com/paper/1711.04598/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Nov 2017.

## Built on by (most cited first, 9 of 9)

- [Facial expression and attributes recognition based on multi-task learning of lightweight neural networks](https://mlsanity.com/paper/2103.17107/) (2021)
- [Expression Snippet Transformer for Robust Video-based Facial Expression Recognition](https://mlsanity.com/paper/2109.08409/) (2021)
- [Pre-training strategies and datasets for facial representation learning](https://mlsanity.com/paper/2103.16554/) (2021)
- [Pose-disentangled Contrastive Learning for Self-supervised Facial Representation](https://mlsanity.com/paper/2211.13490/) (2022)
- [MIDAS: Mixing Ambiguous Data with Soft Labels for Dynamic Facial Expression Recognition](https://mlsanity.com/paper/2503.00202/) (2025)
- [ComFace: Facial Representation Learning with Synthetic Data for Comparing Faces](https://mlsanity.com/paper/2405.16016/) (2024)
- [Enhancing Ambiguous Dynamic Facial Expression Recognition with Soft Label-based Data Augmentation](https://mlsanity.com/paper/2506.20867/) (2025)
- [A Peek at Peak Emotion Recognition](https://mlsanity.com/paper/2205.09791/) (2022)
- [Quality-Controlled Multimodal Emotion Recognition in Conversations with Identity-Based Transfer Learning and MAMBA Fusion](https://mlsanity.com/paper/2511.14969/) (2025)

## Abstract (arXiv)

In this paper we describe a solution to our entry for the emotion recognition challenge EmotiW 2017. We propose an ensemble of several models, which capture spatial and audio features from videos. Spatial features are captured by convolutional neural networks, pretrained on large face recognition datasets. We show that usage of strong industry-level face recognition networks increases the accuracy of emotion recognition. Using our ensemble we improve on the previous best result on the test set by about 1 %, achieving a 60.03 % classification accuracy without any use of visual temporal information.