Review on Deep Face Drawing on basis of Sketch using CNN
Abstract
To catch the culprit based on arbitrary softwareor hand-generated witness sketches, this technique is usefulwhen evidence is scarce. Recent image-to-image deep conversion techniques can rapidly generate facial images from his freehand sketches. However, existing solutions tendto be too sketch- friendly, requiring professionalsketches and even edge maps as input. A key idea to address this problem is to implicitly model a plausible faceimage shape space andsynthesize face images in thisspace to approximate the input sketch. Our method basically uses the inputsketches as soft boundaryconditions, so it can generate high-quality facial images even from rough or imperfect sketches. we developed a CNN framework thatuses CNN algorithms to transform sketch images into realistic human images.
Full Text:
PDFReferences
Guo, Q., Zhu, C., Xia, Z., Wang, Z., & Liu, Y. (2017, September). Attribute-controlled face photo synthesis from simple line drawing. In 2017 IEEE international conference on image processing (ICIP) (pp. 2946-2950). IEEE.
Yeping, F., Desheng, Y., Dong, M., Dian, J., Xiaojin, C., & Mengxian, C. (2019, September). A research on face cognition method with deep ensemble learning and feedback mechanism. In 2019 2nd International Conference on Information Systems and Computer Aided Education (ICISCAE) (pp. 588-593). IEEE.
Sheng, B., Li, P., Gao, C., & Ma, K. L. (2018). Deep neural representation guided face sketch synthesis. IEEE transactions on visualization and computer graphics, 25(12), 3216-3230.
Peng, B., Yang, H., Li, D., & Zhang, Z. (2018, May). An empirical study of face recognition under variations. In 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018) (pp. 310-317). IEEE.
Lum, E. B., & Ma, K. L. (2001, October). Non-photorealistic rendering using watercolor inspired textures and illumination. In Proceedings Ninth Pacific Conference on Computer Graphics and Applications. Pacific Graphics 2001 (pp. 322-330). IEEE.
Refbacks
- There are currently no refbacks.