Yuan Zhao, Xin Tong, Zichong Zhu, Jianda Sheng, Lei Dai, Lingling Xu, Xuehai Xia, Jiang Yu, Jiao Li. 2022. Rethinking Optical Flow Methods for Micro-Expression Spotting: The 30th ACM International Conference on Multimedia Proceedings (MM’22), October 10-14, 2022, Lisboa, Portugal. ACM, New York, NY, USA. 5 pages. https://doi.org/10.1145/3503161.3551602
The branch megc2022 is the source code the results which are submitted to ACMMM 2022 Conference : Grand Challenges.
We also improve the algorithm after the challenge. Some latest results are shown below.
Here shows some results after improve the algorithm, which is better than the results of the code on github.
table: performac on CAS(ME)^2
| iou | 0.2 | 0.3 | 0.4 | 0.5 |
|---|---|---|---|---|
| total precision | 0.4551 | 0.4342 | 0.4175 | 0.3862 |
| total recall | 0.6140 | 0.5859 | 0.5633 | 0.5211 |
| total F1-score | 0.5227 | 0.4988 | 0.4796 | 0.4436 |
| MaE precision | 0.4682 | 0.4470 | 0.4305 | 0.4023 |
| MaE recall | 0.6633 | 0.6333 | 0.6100 | 0.5700 |
| MaE F1-score | 0.5489 | 0.5241 | 0.5048 | 0.4717 |
| ME precision | 0.3518 | 0.3333 | 0.3148 | 0.2592 |
| ME recall | 0.3333 | 0.3157 | 0.2982 | 0.2456 |
| ME F1-score | 0.3423 | 0.3243 | 0.3063 | 0.2522 |
fig shows the prediction of video 16_0101disgustingteeth in CAS(ME)^2. Boxes with filled color are the ground truth, and blank is the predictions.
Video shows the routing of predictions.
Our article was inspired by the work of HE Yuhong (Research on Micro-Expression Spotting Method Based on Optical Flow Features), and the code can be downloaded here.
[1] Qu, F., Wang, S. J., Yan, W. J., Li, H., Wu, S., & Fu, X. (2017). CAS(ME)2: a database for spontaneous macro-expression and micro-expression spotting and recognition. IEEE Transactions on Affective Computing, 9(4), 424-436.