Xiaohan Jin (Sharon)

August 01, 2017

Based on Show-and-Tell image captioning model,  we improved the performance by adding audio features, spatial attention, image tags, and GAN loss for video captioning.

May 15, 2017

 Achieved real-time video style transfer based on Markovian GAN and improved flow consistency by DeepFlow.

August 01, 2017

I worked on core perception part Carnegie Mellon team in Amazon Robotics Challenge 2017.

January 13, 2016

I realized SLAM and sensor fusion on wheeled robots.

See how I made a robot reconstruct the world!

April 08, 2014

Created novel panoramic unwrapping solution based on salient detection and image segmentation.

Researched automatic rectification method for panoramic images. 

I learnt how to do scientific research in a sea of papers and conferences.

This is my first publication. Find me in Siggraph Asia 2015!

October 01, 2015

I rendered some special 3D visual effects by modifying convergence plane in a stereoscopic system according to attention and psychological horopter consistency.

October 22, 2015

I worked on autonomous driving @Deep Glint Co., Ltd., a leading AI company in China.

I have great passion in vision-based navigation for driverless cars.

June 15, 2016

I am cooperating with Microsoft Research Asia and Adminitration of Longmen Grottoes of World Cultural Heritage! 

It's a great opportunity to combine the state-of-the-art computer vision techniques with heritage protection!

August 19, 2015

I worked on dense 3D reconstruction by stereo cameras, object tracking, and virtual reality display for surveillance system during my internship @DeepGlint Co., Ltd.

June 15, 2016

I took an internship in a leading computer vision startup - SenseTime,  whose face recognition techniques (DeepID) rank top 5 in ImageNet Competition.