Summary of Compound Expression Recognition Via Multi Model Ensemble, by Jun Yu et al.
Compound Expression Recognition via Multi Model Ensemble
by Jun Yu, Jichao Zhu, Wangyuan Zhu
First submitted to arxiv on: 19 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Compound Expression Recognition (CER) is a vital component in understanding human emotional expressions, which can be complex due to the presence of compound expressions. To address this challenge, we propose an ensemble learning-based approach for CER. Specifically, we develop three expression classification models using convolutional networks, Vision Transformers, and multi-scale local attention networks. We then employ late fusion to merge the outputs of these models, enabling accurate predictions through model ensemble. Our proposed method demonstrates high accuracy on the RAF-DB dataset and can recognize expressions through zero-shot learning on certain portions of C-EXPR-DB. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about recognizing human emotions from facial expressions. It’s hard to understand people’s feelings because there are many different expressions that can mean the same thing. To solve this problem, scientists developed a new way to recognize compound expressions using artificial intelligence. They created three models that look at faces in different ways and then combined their results to get a more accurate answer. This method works well on some datasets and can even recognize emotions without training on all of them. |
Keywords
» Artificial intelligence » Attention » Cer » Classification » Zero shot
