Search

Facial micro-expression datasets lack consistency and standardisation, with different research groups using various experimental settings, in particular, where the datasets are varied in resolution and frame rates. To provide new insights into the roles of frame rate and resolution, we conduct an investigation into the use of different frame rates and resolution on current benchmark datasets (SMIC and CASME II). By using Temporal Interpolation Model, we subsample SMIC (original frame rate is 100 fps) to 50 fps and CASME II (original frame rate is 200 fps) into 100 fps and 50 fps. In addition, the resolution settings are adjusted to three scaling factors: 100% (original resolution), 75% and 50%. Three feature types are used to test the performance of these settings, which are Local Binary Patterns in Three Orthogonal Planes, 3D Histograms of Oriented Gradient and Histogram of Oriented Optical Flow. The results showed that the frame rate and resolution could affect the performance of micro-expression recognition, which behave distinctively dependent on feature types. This work provides new guidelines for future research in selecting frame rate, resolution and feature descriptors in micro-expressions recognition.

This paper presents baseline results for the first Facial Micro-expressions Grand Challenge (MEGC) 2018 by evaluating LBP-TOP, HOOF and 3DHOG on CASME II and SAMM. We further improve the result of composite database evaluation (Task B of the challenge) by introducing selective block-based features fusion representation. Base on objective classes, this task combines CASME II and SAMM into a single composite database and uses Leave-One-Subject-Out crossvalidation to evaluate the performance. Our proposed method achieve F1-Score of 0.579, which outperformed LBP-TOP, HOOF and 3DHOG with 0.523, 0.527 and 0.436, respectively.

Micro-facial expressions are regarded as an important human behavioural event that can highlight emotional deception. Spotting these movements is difficult for humans and machines, however research into using computer vision to detect subtle facial expressions is growing in popularity. This paper proposes an individualised baseline micro-movement detection method using 3D Histogram of Oriented Gradients (3D HOG) temporal difference method. We define a face template consisting of 26 regions based on the Facial Action Coding System (FACS). We extract the temporal features of each region using 3D HOG. Then, we use Chi-square distance to find subtle facial motion in the local regions. Finally, an automatic peak detector is used to detect micro-movements above the newly proposed adaptive baseline threshold. The performance is validated on two FACS coded datasets: SAMM and CASME II. This objective method focuses on the movement of the 26 face regions. When comparing with the ground truth, the best result was an AUC of 0.7512 and 0.7261 on SAMM and CASME II, respectively. The results show that 3D HOG outperformed for micro-movement detection, compared to state-of-the-art feature representations: Local Binary Patterns in Three Orthogonal Planes and Histograms of Oriented Optical Flow.

Micro-expressions are brief spontaneous facial expressions that appear on a face when a person conceals an emotion, making them different to normal facial expressions in subtlety and duration. Currently, emotion classes within the CASME II dataset are based on Action Units and self-reports, creating conflicts during machine learning training. We will show that classifying expressions using Action Units, instead of predicted emotion, removes the potential bias of human reporting. The proposed classes are tested using LBP-TOP, HOOF and HOG 3D feature descriptors. The experiments are evaluated on two benchmark FACS coded datasets: CASME II and SAMM. The best result achieves 86.35\% accuracy when classifying the proposed 5 classes on CASME II using HOG 3D, outperforming the result of the state-of-the-art 5-class emotional-based classification in CASME II. Results indicate that classification based on Action Units provides an objective method to improve micro-expression recognition., Comment: 11 pages, 4 figures and 5 tables. This paper will be submitted for journal review