no code implementations • 29 Jan 2022 • Yunfang Fu, Qiuqi Ruan, Ziyan Luo, Gaoyun An, Yi Jin, Jun Wan
In this paper, a novel approach via embedded tensor manifold regularization for 2D+3D facial expression recognition (FERETMR) is proposed.
3D Facial Expression Recognition Dimensionality Reduction +1
1 code implementation • 6 Apr 2021 • Xin Wang, Yang Zhao, Tangwen Yang, Qiuqi Ruan
In this paper, we propose a multi-scale context aggregation network (MSCANet) based on single-column encoder-decoder architecture for crowd counting, which consists of an encoder based on a dense context-aware module (DCAM) and a hierarchical attention-guided decoder.
1 code implementation • 9 May 2019 • Jieru Jia, Qiuqi Ruan, Timothy M. Hospedales
Specifically, we observe that the domain discrepancy in \reid{} is due to style and content variance across datasets and demonstrate appropriate Instance and Feature Normalization alleviates much of the resulting domain-shift in Deep \reid{} models.
no code implementations • 22 Nov 2017 • Zhenxing Zheng, Gaoyun An, Qiuqi Ruan
Each stream consists of a Residual Networks(ResNets) and a recurrent model.