Changqun Xia

Currently I am an associate professor in Peng Cheng Laboratory (PCL), Shenzhen, China.

I have received the Ph.D degree from the State Key Laboratory of Virtual Reality Technology and Systems, Beihang University, advised by Prof. Jia Li. I am the author of refereed journals and conferences such as TPAMI, TIP, CVPR and ICCV. Email: xiachq@pcl.ac.cn.

Email  /  Google Scholar  /  CVTeam

I study computer vision and machine learning. My research lies much in image and video processing with learning and optimization methods.

We explore that it is difficult to segment large or small-scale objects due to their asymmetric segmentation requirements. We deconstruct the role of receptive fields in SOD and introduce a Bilateral Extreme Stripping encoder based on the simplified vision transformer and the lightweight CNN for the broader receptive fields.

We construct a 360 omnidirectional image-based SOD dataset, namely ODI-SOD. It has object-level pixelwise annotations on ERP images and is the largest dataset for 360 ISOD by far to our best knowledge. Moreover, we propose a view-aware salient object detection method for 360 ODIs.

We contribute a new Ultra-High-Resolution Saliency Detection dataset UHRSD, containing 5920 images at 4K-8K resolutions. To our knowledge, it is the largest dataset in both quantity and resolution for high-resolution SOD task. We propose a novel one-stage framework called Pyramid Grafting Network.

We rethink the two difficulties that hinder the development of salient object detection, which consists of indistinguishable regions and complex structures. To solve these two issues, we propose the purificatory network with structural similarity loss.

Existing methods usually aggregate the low-level features containing details and the high-level features containing semantics over a large span, which introduces noise into the aggregated features and generates inaccurate saliency maps. In this paper, we propose a pyramidal feature shrinking network.

Most of existing SOD methods focus more on performance than efficiency. Besides, the U-shape structure exists some drawbacks and there is still a lot of room for improvement. Therefore, we propose a novel framework to treat semantic context, spatial detail and boundary information separately in the decoder part.

Task-aware SOD has hardly been studied due to the lack of task-specific datasets. In this paper, we construct a driving task-oriented dataset where pixel-level masks of salient objects have been annotated. We proposed a baseline model for the driving task-aware SOD via a knowledge transfer convolutional neural network.

SOD on 360 omnidirectional images is less studied owing to the lack of datasets with pixel-level annotations. Toward this end, this paper proposes a 360 image-based SOD dataset that contains 500 high-resolution equirectangular images. We proposes a baseline model for SOD on equirectangular images to deal with the distortion caused by the equirectangular projection.

Typically, a salient object detection (SOD) model faces opposite requirements in processing object interiors and boundaries. To address this selectivityinvariance dilemma, we propose a novel boundary-aware network with successive dilation for image-based SOD.

Extended version of CVPR 2015 with improved network flow solver and object shape prior.

video-based SOD is much less explored due to the lack of large-scale video datasets within which salient objects are unambiguously defined and annotated. Toward this end, this paper proposes a video-based SOD dataset that consists of 200 videos. Based on this dataset, this paper proposes an unsupervised baseline approach for video-based SOD by using saliency-guided stacked autoencoders.

Finding what is and what is not a salient object can be helpful in developing better features and models in salient object detection (SOD). In this paper, we investigate the images that are selected and discarded in constructing a new SOD dataset and find that many similar candidates, complex shape and low objectness are three main attributes of many non-salient objects. Then we propose a novel salient object detector by ensembling linear exemplar regressors.

Weak object detectors can generate strong video object segmentation results via joint inference with a quadratic network flow model.

Existing metrics, which are often heuristically designed, may draw conflict conclusions in comparing saliency models. As a consequence, it becomes somehow confusing on the selection of metrics in comparing new models with state-of-the-arts. To address this problem, we propose a data-driven metric for comprehensive evaluation of saliency models.

We propose an adaptive template for semantic labeling of indoor scene objects and estimating their oriented bounding facets (OBFs). The proposed adaptive template encodes prior geometric information of objects based on statistics of the training images.

Much thanks to Jon Barron for sharing this template.