ABSTRACT
As the first two steps of face recognition system, face detection and key point location are of great significance to the accuracy of face recognition system. The Multi-task Cascaded Convolutional Network(MTCNN) uses a single network model for the first time to complete the task of face detection and key point location. It is easy to find the problem of false detection in the practice of people's face detection. Therefore, this paper introduces multi-scale and multi template data preprocessing in MTCNN model. By increasing the diversity of data, the accuracy of MTCNN model is enhanced. The results show that compared with the original MTCNN model, the recall of the hybrid test set composed of WIDER Face and Celeba is increased by 2% after introducing multi-scale and multi template data preprocessing.
- Milborrow S, Nicolls F. Locating facial features with an extended active shape model[C]. European conference on computer vision. Springer, Berlin, Heidelberg, 2008: 504-513.Google ScholarDigital Library
- Wan Kwok-Wai, Lam K M, Ng K C. An accurate active shape model for facial feature extraction [J]. Pattern recognition letters, 2005, 26 (15): 2409-2423.Google Scholar
- Cootes T F, Edwards G J, Taylor C J. Active appearance models [J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2001 (6): 681- 685Google ScholarDigital Library
- Kahraman F, Gokmen M, Darkner S, An active illumination and appearance (AIA) model for face alignment[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2007: 1-7.Google Scholar
- Burgos-Artizzu X P, Perona P, Dollár P. Robust face landmark estimation under occlusion[C]. IEEE International Conference on Computer Vision, 2013: 1513-1520Google Scholar
- Chen Dong, Ren Shaoqing, Wei Yichen, Joint cascade face detection and alignment[C]. European Conference on Computer Vision. Springer, Cham, 2014: 109-122Google Scholar
- SUN Y, WANG X G, TANG X O. Deep convolutional network cascade for facial point detection[C]. IEEE conference on computer vision and pattern recognition, 2013: 3476 − 3483.Google Scholar
- ZHOU E J, FAN H Q, CAO Z M, Extensive facial landmark localization with coarse-to-fine convolutional network cascade[C]. IEEE international conference on computer vision, 2013: 386 – 391.Google Scholar
- Zhang, Kaipeng, Joint face detection and alignment using multitask cascaded convolution function and GLCM. Autex Research Journal, 2015: 226-232.Google Scholar
- S. Yang, P. Luo, C. C. Loy, and X. Tang. WIDER FACE: A Face Detection Benchmark. arXiv preprint arXiv:1511.06523Google Scholar
- M.G.P. Rose, S.M. Palmer, and X. Tang. Deep learning face attributes in the world[C]. IEEE International Conference on Computer Vision, 2015: 3730-3738Google Scholar
- Y. Sun, X. Wang, and X. Tang. Deep Convolutional NetWork Cascade for Facial Point Detection[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2013: 2570-2576Google Scholar
- GROSSR, MATTHEWSI, COHNJ, etal. Multi-PIE[J]. Image and Vision Computing, 2010, 28(5), 807-813Google ScholarDigital Library
- Ketkar N.Introduction to PyTorch [M]. Deep Learning with Python. 2017Google Scholar
- Lv Jiangjing, Shao Xiaohu, Xing Junliang, A deep regression architecture with two-stage re-initialization for high performance facial landmark detection[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2017: 3317-3326Google Scholar
Recommendations
Using DSCB: A Depthwise Separable Convolution Block Rebuild MTCNN for Face Detection
ICIGP '22: Proceedings of the 2022 5th International Conference on Image and Graphics ProcessingNowadays, there are huge demands of face detection in images and videos for surveillance, education, autonomous driving and health care. These application scenarios need high accuracy and efficiency of face detection. However, in some scene, ...
MTCNN and FACENET Based Access Control System for Face Detection and Recognition
AbstractFace detection and recognition is one of the research hotspots in the field of computer vision, which is widely used in video surveillance and identity matching. The traditional algorithms of face detection include AdaBoost, Haar-like, DPM, etc. ...
Head Pose Estimation via Multi-Task Cascade CNN
HPCCT '19: Proceedings of the 2019 3rd High Performance Computing and Cluster Technologies ConferenceIn our daily life, many face applications need to complete three tasks: face detection, facial landmark localization and head pose estimation. Currently, most methods accomplish these three tasks separately. Multi-task cascade convolution neural network(...
Comments