Micro-and-Macro Expression Warehouse (MMEW) for Micro-expression Recognition
April 1, 2026 · View on GitHub
MMEW dataset is pulished in paper "Video-based Facial Micro-Expression Analysis: A Survey of Datasets, Features and Algorithms"[1].
Introduction
MMEW includes 300 micro-expressions and 900 macro-expressions which include six basic types, i.e. happiness, surprise, anger, disgust, fear and sadness. For convenient evaluation, all the samples are well named. Micro-expression sequence images named as “frameIndex.jpg” are stored in a three-layer file directory. Specifically, subject indexes are used as root folder, then some sub-folders, i. e., happiness, surprise, disgust, repression, fear, sadness, anger, each of which includes micro-expression sequence images. To better understand which image belongs to which subject and which emotion, the third layer directory is named in the way of “PersonIndex-EmotionIndex-SampleIndex”. For instance, an image stored in the path “S03/happiness/S03-01-002/1.jpg” is the 1st frame of the 2nd micro-expression sequence with happiness emotion for the 3rd subject. Similarly, the images for macro-expression are stored in a three-layer file directory. The first directory embodies a subject index, and the second directory embodies a macro-expression emotion. For example, “S03/happiness/S03-01-001.jpg” means the 1st sample with happiness emotion for the 3rd subject.
For more information, please see the introduction video or the website of our lab here.
How to get
MMEW dataset is publicly available for any academic use. To obtain MMEW, please sign the license agreement and send a scanned copy to benxianyeye@163.com.
Reference
[1] Xianye Ben, Yi Ren, Junping Zhang, Su-Jing Wang*, Kidiyo Kpalma, Weixiao Meng, Yong-Jin Liu*. Video-based Facial Micro-Expression Analysis: A Survey of Datasets, Features and Algorithms. IEEE Transactions on Pattern Analysis and Machine Intelligence, DOI: 10.1109/TPAMI.2021.3067464