Summary on Facial Landmark Detection
Available Online June 2016.
- https://doi.org/10.2991/mecs-17.2017.47How to use a DOI?
- Facial Landmark Detection, DCNN, HeHOP, partition input domain, PIMV.
- Facial landmark detection has important applications in many aspects such as facial recognition, expression recognition, facial attributes analysis, and so on. It compares the detected images with images in dataset to find matched faces, which achieves the identification purpose. Because of its wide use in business, security, identification and other aspects, it gains more and more attention. State-of-art researches on deep convolutional neutral network (DCNN) separate DCNN into DCNN-I(Inner) and DCNN-C(Contour) to get more accurate detection. In HeHOP estimation, performed on local areas projected relative to the head orientation and position, a binary feature extraction approach based on depth data is developed to estimate the head orientation directly by global linear regression. An energy based partitioning of the input domain with a direct control of the final variance perpartition is proposed. A pose-indexed based multi-view (PIMV) face alignment method that obtains a more accurate prediction is proposed in. Researches also proposed a method to handle both images with severe occlusion and images with large head poses. In this paper, we will summarize some state-of-art methods on FLD and analysis their advantages and disadvantages. Based on this, we will have a discussion on some feasible research areas.
- Open Access
- This is an open access article distributed under the CC BY-NC license.
Cite this article
TY - CONF AU - Jinghao Wen PY - 2016/06 DA - 2016/06 TI - Summary on Facial Landmark Detection BT - Proceedings of the 2017 2nd International Conference on Machinery, Electronics and Control Simulation (MECS 2017) PB - Atlantis Press SN - 2352-5401 UR - https://doi.org/10.2991/mecs-17.2017.47 DO - https://doi.org/10.2991/mecs-17.2017.47 ID - Wen2016/06 ER -