Synthesizing Imagined Faces Based on Relevance Feedback
In this paper, we propose a user-friendly system that can create a facial image from a corresponding image in the user’s mind. Unlike most of the existing methods, which require a sketch as input or the tedious work of selecting similar facial components from an example database, our method can synthesise a satisfying result without questioning the user on the explicit features of the face in his or her mind. Through a dialogic approach based on a relevance feedback strategy to translate facial features into input, the user only needs to look at several candidate face images and judge whether each image resembles the face that he or she is imagining. A set of sample face images that are based on users’ feedbacks are used to dynamically train an Optimum-Path Forest algorithm to classify the relevance of face images. Based on the trained Optimum-Path Forest classifier, candidate face images that best reflect the user’s feedback are retrieved and interpolated to synthesise new face images that are similar to those the user had imagined. The experimental results show that the proposed technique succeeded in generating images resembling a face a user had imagined or memorised.
KeywordsFace image synthesis Relevance feedback Optimum-Path Forest
This work was supported by JSPS KAKENHI (Grant No. 17H00737) and the Public Projects of Zhejiang Natural Science Foundation Province, China (Grant No. LGF18F020015).
- 1.E-FIT. http://www.visionmetric.com/
- 3.Wu, D., Dai, Q.: Sketch realizing: lifelike portrait synthesis from sketch. In: Computer Graphics International Conference, pp. 13–20 (2009)Google Scholar
- 7.Zhu, S., Liu, S., Loy, C., Tang, X.: Deep cascaded bi-network for face hallucination. In: Computer Vision and Pattern Recognition, pp. 614–630 (2016)Google Scholar
- 10.Wang, B., Zhang, X., Li, N.: Relevance feedback technique for content-based image retrieval using neural network learning. In: International Conference on Machine Learning and Cybernetics (2006)Google Scholar
- 11.Fu, H., Qiu, G.: Fast semantic image retrieval based on random forest. In: International Conference on Multimedia, pp. 909–912 (2012)Google Scholar
- 12.Sun, Y., Chen, Y., Wang, X., Tang, X.: Deep learning face representation by joint identification-verification. In: Advances in Neural Information Processing Systems, pp. 1988–1996 (2014)Google Scholar
- 13.Sun, Y., Wang, X., Tang, X.: Deep learning face representation from predicting 10,000 classes. In: Computer Vision and Pattern Recognition, pp. 1891–1898 (2014)Google Scholar
- 16.Silva, A., Falcao, A., Magalhaes, L.: Active learning paradigms for CRIR systems based on optimum-path forest classification. J. WSCG 18(1–3), 73–80 (2010)Google Scholar
- 17.Papa, J., Falca, A.: Optimum-path forest: a novel and powerful framework for supervised graph-based pattern recognition techniques, pp. 41–48. Institute of Computing University of Campinas (2010)Google Scholar