Display options
Cite
Kim T, Mok J, Lee E. Detecting Facial Region and Landmarks at Once via Deep Network. Sensors (Basel). 2021;21(16)doi: 10.3390/s21165360.
Kim, T., Mok, J., & Lee, E. (2021). Detecting Facial Region and Landmarks at Once via Deep Network. Sensors (Basel, Switzerland), 21(16), . https://doi.org/10.3390/s21165360
Kim, Taehyung, et al. "Detecting Facial Region and Landmarks at Once via Deep Network." Sensors (Basel, Switzerland) vol. 21,16 (2021). doi: https://doi.org/10.3390/s21165360
Kim T, Mok J, Lee E. Detecting Facial Region and Landmarks at Once via Deep Network. Sensors (Basel). 2021 Aug 09;21(16). doi: 10.3390/s21165360. PMID: 34450804; PMCID: PMC8401714.
Copy Download .nbib Format: NLM
Share it on

Sensors (Basel). 2021 Aug 09;21(16). doi: 10.3390/s21165360.

Detecting Facial Region and Landmarks at Once via Deep Network.

Sensors (Basel, Switzerland)

Taehyung Kim, Jiwon Mok, Euichul Lee

Affiliations

  1. Department of AI & Informatics, Graduate School, Sangmyung University, Seoul 03016, Korea.
  2. Department of Human-Centered Artificial Intelligence, Sangmyung University, Seoul 03016, Korea.

PMID: 34450804 PMCID: PMC8401714 DOI: 10.3390/s21165360

Abstract

For accurate and fast detection of facial landmarks, we propose a new facial landmark detection method. Previous facial landmark detection models generally perform a face detection step before landmark detection. This greatly affects landmark detection performance depending on which face detection model is used. Therefore, we propose a model that can simultaneously detect a face region and a landmark without performing the face detection step before landmark detection. The proposed single-shot detection model is based on the framework of YOLOv3, a one-stage object detection method, and the loss function and structure are altered to learn faces and landmarks at the same time. In addition, EfficientNet-B0 was utilized as the backbone network to increase processing speed and accuracy. The learned database used 300W-LP with 64 facial landmarks. The average normalized error of the proposed model was 2.32 pixels. The processing time per frame was about 15 milliseconds, and the average precision of face detection was about 99%. As a result of the evaluation, it was confirmed that the single-shot detection model has better performance and speed than the previous methods. In addition, as a result of using the COFW database, which has 29 landmarks instead of 64 to verify the proposed method, the average normalization error was 2.56 pixels, which was also confirmed to show promising performance.

Keywords: deep network; face detection; face landmark detection; simultaneous detection; single-shot detection

References

  1. IEEE Trans Pattern Anal Mach Intell. 2006 Dec;28(12):2037-41 - PubMed
  2. IEEE Trans Pattern Anal Mach Intell. 2011 Feb;33(2):394-405 - PubMed
  3. IEEE Trans Pattern Anal Mach Intell. 2013 Dec;35(12):2930-40 - PubMed

MeSH terms

Publication Types

Grant support