Bing Shuai's homepage

Hello, My name is Bing, Welcome to my Google site.

I have recently joined Amazon as a research scientist in Seattle, United States. Prior to that, I finished my PhD studies in Nanyang Technological University, Singapore. There, I was under the supervision of professor Kim-Hui Yap and professor Gang Wang, who recently moved to Alibaba AI Lab. I was always very grateful to their support and encouragement during the hard journey. My PhD thesis focuses on exploring deep learning for scene segmentation task. In the mean time, I'm always interested in developing efficient algorithms to perform accurate scene analysis. 

Now, I'm also working closely with Associate Professor Jianfang Hu (Sun Yat-sen University, China) to extend research to fast and video segmentation task.

I also spent half-year as a research intern in Panasonic Singapore starting from Aug-2017, where I explored to develop efficient object detection algorithms for real-world applications. I also met many interesting people there. Thank you, everyone!!!

Professional FilesLinkedin, Google Scholar

Feel free to reach out via my google mail: beinshuai AT gmail.com.

Publications

Jiafeng Xie, Bing Shuai, Jianfang Hu, Jingyang Lin, Weishi Zheng. Improving Fast Segmentation Network with Teacher-Student Learning. British Machine Vision Conference (BMVC) 2018. (New!!!)

Heng-Hui Ding,  Xudong Jiang, Bing Shuai, Ai Qun Liu, Gang Wang. Context Contrasted Feature and Gated Multi-scale Aggregation for Scene SegmentationIEEE Proceedings on Computer Vision and Pattern Recognition (CVPR) 2018. (Oral)

Bing Shuai, Zhen Zuo, Bing Wang, Gang Wang. Scene Segmentation with DAG-Recurrent Neural Networks. IEEE Transaction on Pattern Analysis and Machine Intelligence (TPAMI), 2017. 

Abrar Abul Nabi , Bing Shuai, Gang Wang. Episodic CAMN: Contextual Attention-based Memory Networks for Scene Labeling. IEEE Proceedings on Computer Vision and Pattern Recognition (CVPR), 2017.

Ping Hu, Bing Shuai, Jun Liu, Gang Wang. Deep Level Sets for Salient Object Detection. IEEE Proceedings on Computer Vision and Pattern Recognition (CVPR), 2017.

Rahul R.Varior, Bing Shuai, Jiwen Lu, Dong Xu, Gang Wang.  A siamese long short-term memory architecture for human re-identication. European Conference on Computer Vision (ECCV), 2016.

Bing Shuai, Zhen Zuo, Gang Wang, Bing Wang. DAG-Recurrent Neural Networks For Scene Labeling.  IEEE Proceedings on Computer Vision and Pattern Recognition (CVPR), 2016.[PDF] [DAG-RNN(4) Code]

Bing Shuai, Gang Wang, Zhen Zuo, Bing Wang, Lifan Zhao. Integrating Parametric and Non-Parametric Models for Scene Labeling. IEEE Proceedings on Computer Vision and Pattern Recognition (CVPR), 2015.  [PDF]. [CNN Training Code]. [CNN Model]. [Global Labeling Code]

Bing Shuai, Zhen Zuo, Gang Wang, Bing Wang. Scene Parsing with Integration of Parametric and Non-parametric Models. IEEE Transaction on Image Processing (TIP) 2016. [PDF]

Bing Shuai, Zhen Zuo, Gang Wang; Quaddirectional 2D-Recurrent Neural Networks For Image Labeling.  IEEE Signal Processing Letters 2016 [PDF]

Zhen Zuo, Bing Shuai, Gang Wang, Xiao Liu, Xingxing Wang, Bing Wang. Convolutional Recurrent Neural Networks: Learning Spatial Dependencies for Image Representation. IEEE Proceedings on Computer Vision and Pattern Recognition  Workshop (CVPRW), 2015. [PDF

Zhen Zuo, Bing Shuai, Gang Wang, Xiao Liu, Xingxing Wang, Bing Wang, Yushi Chen. Learning Contextual Dependencies with Convolutional Hierarchical Recurrent Neural Networks .  IEEE Transaction on Image Processing (TIP) 2016 [PDF]

Zhen Zuo, Gang Wang, Bing Shuai, Lifan Zhao, Qingxiong Yang, Xudong Jiang. Learning Discriminative and Shareable Features for Scene Classification. European Conference on Computer Vision (ECCV) 2014 [PDF]

Zhen Zuo, Gang Wang, Bing Shuai, Lifan Zhao, Qingxiong Yang, Xudong Jiang; Exemplar Based Deep Discriminative and Shareable Feature Learning for Scene Image Classification.  Pattern Recognition 2015 [PDF]


Pre-prints



Bing Shuai, Ting Liu, Gang Wang; Improving Fully Convolution Network for Semantic Segmentation. [PDF] [Supplementary Material]

The preliminary model (significantly inferior to the published model) achieves 6-th place (out of 23 teams) in ImageNet Scene Parsing 2016 Challenges. (Note that no post-processing or auxiliary labels are used to enhance the predictions)



Datasets

Bing Wang, Kap Luk Chan, Li Wang, Bing Shuai, Zhen Zuo, Ting Liu, Gang Wang; Joint Learning of Convolutional Neural Networks and Temporally Constrained Metrics for Tracklet Association Based on Large-Scale Datasets. [Arxiv preprint] [Training Data] [Testing Data].

The data format of this new dataset follows the MOTchallengeToolbox in MOTchallenge may be used to parse the groundtruth data.