K
Kun Zhao
Researcher at University of Queensland
Publications - 38
Citations - 289
Kun Zhao is an academic researcher from University of Queensland. The author has contributed to research in topics: Manifold & Computer science. The author has an hindex of 7, co-authored 33 publications receiving 163 citations. Previous affiliations of Kun Zhao include NICTA.
Papers
More filters
Journal ArticleDOI
Faster ILOD: Incremental learning for object detectors based on faster RCNN
TL;DR: This paper designs an efficient end-to-end incremental object detector using knowledge distillation for object detectors based on RPNs and introduces multi-network adaptive distillation that properly retains knowledge from the old categories when fine-turning the model for new task.
Proceedings ArticleDOI
SOS: Selective Objective Switch for Rapid Immunofluorescence Whole Slide Image Classification
TL;DR: It is demonstrated that conventional patch-based processing is redundant for certain WSI classification tasks where high resolution is only required in a minority of cases and a method is proposed for the selective use of high resolution processing based on the confidence of predictions on downscaled WSIs, called the Selective Objective Switch (SOS).
Journal ArticleDOI
Efficient clustering on Riemannian manifolds
TL;DR: This work proposes a kernelised random projection framework for clustering manifold points via kernel space, which can preserve the geometric structure of the original space, but is computationally efficient.
Posted Content
Efficient Clustering on Riemannian Manifolds: A Kernelised Random Projection Approach
TL;DR: In this paper, the authors propose an efficient framework to address the clustering problem on Riemannian manifolds, which implements random projections for manifold points via kernel space, which can preserve the geometric structure of the original space, but is computationally efficient.
Proceedings ArticleDOI
Random projections on manifolds of Symmetric Positive Definite matrices for image classification
TL;DR: Experiments on face recognition, person re-identification and texture classification show that the proposed approach outperforms several recent methods, such as Tensor Sparse Coding, Histogram Plus Epitome, Riemannian Locality Preserving Projection and Relational Divergence Classification.