Adobe Research Internship
Salient Object Detection for Images using Deep Learning.
Mitacs Research Internship at University of Victoria
Devise an algorithm to alleviate the server bandwidth consumption in a P2P VoD system.
Bachelor Degree at Dalian University of Technology
Major in Electronic and Information Engineering.
How to pronounce my name Hongyang? It's Home
I am currently a 3rd year (2017.8 - 2018.7) PhD student at The Chinese University of Hong Kong under supervision of Prof. Xiaogang Wang
My research covers a wide span of Deep Learning and its applications in Computer Vision. In particular, I am interested in the fully end-to-end learning with different types of models (CNN, RNN, etc) in object detection. Recently I am more interested in using novel Deep RL algorithms to solve various vision tasks.
Rethinking Feature Discrimination and Polymerization for Large-scale Recognition
How to train a deep network to acquire discriminative features across categories and polymerized features within classes has always been at the core of many computer vision tasks. In this paper, we address this problem based on the simple intuition that the cosine distance of features in high-dimensional space should be close enough within one class and far away across categories. To this end, we proposed the congenerous cosine (COCO) algorithm to simultaneously optimize the cosine similarity among data. It inherits the softmax property to make inter-class features discriminative as well as shares the idea of class centroid in metric learning.
Yu Liu*, Hongyang Li*, Xiaogang Wang (* equal contribution)
Recurrent Scale Approximation for Object Detection in CNN
Since CNN lacks an inherent
mechanism to handle large scale variations, we always
need to compute feature maps multiple times for multiscale
object detection. To address this, we devise a recurrent
scale approximation to compute feature map
once only, and only through this map can we approximate
the rest maps on other levels. To further increase efficiency
and accuracy, we (a): design a scale-forecast network
to globally predict potential scales in the image since
there is no need to compute maps on all levels.
(b): propose a landmark retracing network to
retrace back locations of the regressed landmarks and generate
a confidence score for each landmark.
Yu Liu, Hongyang Li, Junjie Yan, F. Wei, Xiaogang Wang, X. Tang
Zoom Out-and-In Network with Recursive Training for Object Proposal
We propose a zoom-out-and-in network for
generating object proposals. We utilize different resolutions
of feature maps in the network to detect object instances of
various sizes. Deeper feature maps contain
region-level semantics which can help shallow counterparts
to identify small objects. Therefore we design a
zoom-in sub-network to increase the resolution of high level
features via a deconvolution operation. Furthermore, we
devise a recursive training pipeline to consecutively regress
region proposals at the training stage in order to match the
iterative regression at the testing stage.
Hongyang Li, Yu Liu, Wanli Ouyang, Xiaogang Wang
arXiv preprint 2017
Dual Deep Network for Visual Tracking
In this paper, we propose a dual network to better utilize features among layers for visual tracking. It is observed that features in higher layers encode semantic context while its counterparts in lower layers are sensitive to discriminative appearance. Thus we exploit the hierarchical features in different layers of a deep model and design a dual structure to obtain better feature representation from various streams, which is rarely investigated in previous work. To leverage the robustness of our dual network, we train it with random patches measuring the similarities between the network activation and target appearance. Quantitative and qualitative evaluations on two large-scale benchmark data sets show that the proposed algorithm performs favourably against the state-of-the-arts.
Hongyang Li, Huchuan Lu, Ming-Hsuan Yang
IEEE Trans. on Image Processing (TIP) 2017
Multi-Bias Non-linear Activation in Deep Neural Networks
In this paper, we propose a multi-bias non-linear
activation (MBA) layer to explore the information
hidden in the magnitudes of responses. It
is placed after the convolution layer to decouple
the responses to a convolution kernel into multiple
maps by multi-thresholding magnitudes, thus
generating more patterns in the feature space at
a low computational cost. It provides great flexibility
of selecting responses to different visual
patterns in different magnitude ranges to form
rich representations in higher layers.
Hongyang Li, Wanli Ouyang, Xiaogang Wang
Learning Deep Representation with Large-scale Attributes
This paper contributes a large-scale object attribute database that contains rich attribute annotations (rotation, viewpoint, occlusion, etc.) for around 180k samples and 494 object classes. We use this dataset to train deep representations and extensively evaluate how these attributes are useful on the general object detection task.
Wanli Ouyang, Hongyang Li, Xingyu Zeng, Xiaogang Wang
Inner and Inter Label Propagation: Salient Object Detection in the Wild
We propose a novel label propagation based method for saliency detection. A key observation is that saliency in an image can be estimated by propagating the labels extracted from the most certain background and object regions.
A co-transduction algorithm is devised to fuse both boundary and objectness labels based on an inter propagation scheme to effectively improve the saliency accuracy.
Hongyang Li, Huchuan Lu, Zhe Lin, Xiaohui Shen, Brian Price
IEEE Trans. on Image Processing (TIP) 2015
My Academic Blog
. A good (probably best) way of sharing ideas than bustling around like most people do in conferences.
Friends and stars
in academia. I 'mark and fork' friends and genii to learn the best out of them. Update:
CV family tree.
Thanks to my Ph.D job, I travel a lot and really enjoy every moment of it. There is a gallery
to take a peek.
Still more unsorted stuff
- Meta talk
: How to give a talk so good that there will be pizza left for you