Shihao Jiang

Shihao Jiang (Zac)

I am a PhD student at Australian National University where I am primarily advised by Richard Hartley. I also work closely with Dylan Campbell and have received guidance from Miaomiao Liu, Hongdong Li and Stephen Gould. My research focuses on optical flow.

Research

	Learning to Estimate Hidden Motions with Global Motion Aggregation Shihao Jiang , Dylan Campbell, Yao Lu, Hongdong Li, Richard Hartley ICCV, 2021 paper / code We show reasoning about self-similarities can help resolve ambiguities caused by occlusions.
	Learning Optical Flow from a Few Matches Shihao Jiang , Yao Lu, Hongdong Li, Richard Hartley CVPR, 2021 paper / code We show accurate optical flow can be predicted from a sparse correlation volume.
	Joint Unsupervised Learning of Optical Flow and Egomotion with Bi-Level Optimization Shihao Jiang , Dylan Campbell, Miaomiao Liu, Stephen Gould, Richard Hartley 3DV, 2020 paper We propose an approach to differentiate through an essential matrix estimation layer and demonstrate its usefulness in unsupervised optical flow estimation.
	Kernel Support Vector Machine and Convolutional Neural Networks Shihao Jiang , Richard Hartley, Basura Fernando DICTA, 2018 paper We embed a kernel support vector machine module in a CNN.