27 citations · 117 across the 22 of their papers we have counts for
22 papers
Training Matting Models without Alpha Labels
Wenze Liu, Zixuan Ye, Hao Lu +2
The labelling difficulty has been a longstanding problem in deep image matting. To escape from fine labels, this work explores using rough annotations such as trimaps coarsely indi…
FADE: A Task-Agnostic Upsampling Operator for Encoder-Decoder Architectures
Hao Lu, Wenze Liu, Hongtao Fu +1
The goal of this work is to develop a task-agnostic feature upsampling operator for dense prediction where the operator is required to facilitate not only region-sensitive tasks li…
In-Context Matting
He Guo, Zixuan Ye, Zhiguo Cao +1
We introduce in-context matting, a novel task setting of image matting. Given a reference image of a certain foreground and guided priors such as points, scribbles, and masks, in-c…
Learning to Upsample by Learning to Sample
Wenze Liu, Hao Lu, Hongtao Fu +1
We present DySample, an ultra-lightweight and effective dynamic upsampler. While impressive performance gains have been witnessed from recent kernel-based dynamic upsamplers such a…
Point-Query Quadtree for Crowd Counting, Localization, and More
Chengxin Liu, Hao Lu, Zhiguo Cao +1
We show that crowd counting can be viewed as a decomposable point querying process. This formulation enables arbitrary points as input and jointly reasons whether the points are cr…
Make-It-4D: Synthesizing a Consistent Long-Term Dynamic Scene Video from a Single Image
Liao Shen, Xingyi Li, Huiqiang Sun +4
We study the problem of synthesizing a long-term dynamic video from only a single image. This is challenging since it requires consistent visual content movements given large camer…
Diffusion-Augmented Depth Prediction with Sparse Annotations
Jiaqi Li, Yiran Wang, Zihao Huang +4
Depth estimation aims to predict dense depth maps. In autonomous driving scenes, sparsity of annotations makes the task challenging. Supervised models produce concave objects due t…
Fast Full-frame Video Stabilization with Iterative Optimization
Weiyue Zhao, Xin Li, Zhan Peng +4
Video stabilization refers to the problem of transforming a shaky video into a visually pleasing one. The question of how to strike a good trade-off between visual quality and comp…
Box-DETR: Understanding and Boxing Conditional Spatial Queries
Wenze Liu, Hao Lu, Yuliang Liu +1
Conditional spatial queries are recently introduced into DEtection TRansformer (DETR) to accelerate convergence. In DAB-DETR, such queries are modulated by the so-called conditiona…
On Point Affiliation in Feature Upsampling
Wenze Liu, Hao Lu, Yuliang Liu +1
We introduce the notion of point affiliation into feature upsampling. By abstracting a feature map into non-overlapped semantic clusters formed by points of identical semantic mean…
NVDS+: Towards Efficient and Versatile Neural Stabilizer for Video Depth Estimation
Yiran Wang, Min Shi, Jiaqi Li +7
Video depth estimation aims to infer temporally consistent depth. One approach is to finetune a single-image model on each video with geometry constraints, which proves inefficient…
Learning Probabilistic Coordinate Fields for Robust Correspondences
Weiyue Zhao, Hao Lu, Xinyi Ye +2
We introduce Probabilistic Coordinate Fields (PCFs), a novel geometric-invariant coordinate representation for image correspondence problems. In contrast to standard Cartesian coor…
A2B: Anchor to Barycentric Coordinate for Robust Correspondence
Weiyue Zhao, Hao Lu, Zhiguo Cao +1
There is a long-standing problem of repeated patterns in correspondence problems, where mismatches frequently occur because of inherent ambiguity. The unique position information a…
Vision Transformer Off-the-Shelf: A Surprising Baseline for Few-Shot Class-Agnostic Counting
Zhicheng Wang, Liwen Xiao, Zhiguo Cao +1
Class-agnostic counting (CAC) aims to count objects of interest from a query image given few exemplars. This task is typically addressed by extracting the features of query image a…
Point-and-Shoot All-in-Focus Photo Synthesis from Smartphone Camera Pair
Xianrui Luo, Juewen Peng, Weiyue Zhao +3
All-in-Focus (AIF) photography is expected to be a commercial selling point for modern smartphones. Standard AIF synthesis requires manual, time-consuming operations such as focal…
Learning Second-Order Attentive Context for Efficient Correspondence Pruning
Xinyi Ye, Weiyue Zhao, Hao Lu +1
Correspondence pruning aims to search consistent correspondences (inliers) from a set of putative correspondences. It is challenging because of the disorganized spatial distributio…
Find Beauty in the Rare: Contrastive Composition Feature Clustering for Nontrivial Cropping Box Regression
Zhiyu Pan, Yinpeng Chen, Jiale Zhang +3
Automatic image cropping algorithms aim to recompose images like human-being photographers by generating the cropping boxes with improved composition quality. Cropping box regressi…
Infusing Definiteness into Randomness: Rethinking Composition Styles for Deep Image Matting
Zixuan Ye, Yutong Dai, Chaoyi Hong +2
We study the composition style in deep image matting, a notion that characterizes a data generation flow on how to exploit limited foregrounds and random backgrounds to form a trai…
SAPA: Similarity-Aware Point Affiliation for Feature Upsampling
Hao Lu, Wenze Liu, Zixuan Ye +3
We introduce point affiliation into feature upsampling, a notion that describes the affiliation of each upsampled point to a semantic cluster formed by local decoder feature points…
Design What You Desire: Icon Generation from Orthogonal Application and Theme Labels
Yinpeng Chen, Zhiyu Pan, Min Shi +3
Generative adversarial networks (GANs) have been trained to be professional artists able to create stunning artworks such as face generation and image style transfer. In this paper…
Robust Object Detection With Inaccurate Bounding Boxes
Chengxin Liu, Kewei Wang, Hao Lu +2
Learning accurate object detectors often requires large-scale training data with precise object bounding boxes. However, labeling such data is expensive and time-consuming. As the…
3D Instances as 1D Kernels
Yizheng Wu, Min Shi, Shuaiyuan Du +3
We introduce a 3D instance representation, termed instance kernels, where instances are represented by one-dimensional vectors that encode the semantic, positional, and shape infor…