Yukang Chen

Yukang Chen

Auto-generated profile, based on publicly available information

Yukang Chen

Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
22 August 2022 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Verified
Journal Article
Closed access
YLYanwei Li
Yanwei Li
HZHengshuang Zhao
Hengshuang Zhao
XQXiaojuan Qi
Xiaojuan Qi
YCYukang Chen
Yukang Chen
LQLu Qi
Lu Qi
JJJiaya Jia
Jiaya Jia
In this paper, we present a conceptually simple, strong, and efficient framework for fully- and weakly-supervised panoptic segmentation, called Panoptic FCN. Our approach aims to represent and predict foreground things and background stuff in a unified fully convolutional pipeline, which can be optimized with point-based fully or weak supervision. In particular, Panoptic FCN encodes each object instance or stuff category with the proposed kernel generator and produces the prediction by convolving the high-resolution feature directly. With this approach, instance-aware and semantically consistent properties for things and stuff can be respectively satisfied in a simple generate-kernel-then-segment workflow. Without extra boxes for localization or instance separation, the proposed approach outperforms the previous box-based and -free models with high efficiency. Furthermore, we propose a new form of point-based annotation for weakly-supervised panoptic segmentation. It only needs several random points for both things and stuff, which dramatically reduces the annotation cost of human. The proposed Panoptic FCN is also proved to have much superior performance in this weakly-supervised setting, which achieves 82% of the fully-supervised performance with only 20 randomly annotated points per instance. Extensive experiments demonstrate the effectiveness and efficiency of Panoptic FCN on COCO, VOC 2012, Cityscapes, and Mapillary Vistas datasets. And it sets up a new leading benchmark for both fully- and weakly-supervised panoptic segmentation. Our code and models are made publicly available at https://github.com/dvlab-research/PanopticFCN.

1 June 2021 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Verified
Journal Article
Closed access
LQLu Qi
Lu Qi
YWYi Wang
Yi Wang
YCYukang Chen
Yukang Chen
YCYing-Cong Chen
Ying-Cong Chen
XZXiangyu Zhang
Xiangyu Zhang
JJJiaya Jia
Jiaya Jia
In this paper, we explore the mask representation in instance segmentation with Point-of-Interest (PoI) features. Differentiating multiple potential instances within a single PoI feature is challenging, because learning a high-dimensional mask feature for each instance using vanilla convolution demands a heavy computing burden. To address this challenge, we propose an instance-aware convolution. It decomposes this mask representation learning task into two tractable modules as instance-aware weights and instance-agnostic features. The former is to parametrize convolution for producing mask features corresponding to different instances, improving mask learning efficiency by avoiding employing several independent convolutions. Meanwhile, the latter serves as mask templates in a single point. Together, instance-aware mask features are computed by convolving the template with dynamic weights, used for the mask prediction. Along with instance-aware convolution, we propose PointINS, a simple and practical instance segmentation approach, building upon dense one-stage detectors. Through extensive experiments, we evaluated the effectiveness of our framework built upon RetinaNet and FCOS. PointINS in ResNet101 backbone achieves a 38.3 mask mean average precision (mAP) on COCO dataset, outperforming existing point-based methods by a large margin. It gives a comparable performance to the region-based Mask R-CNN K. He, G. Gkioxari, P. Dollár, and R. Girshick, “Mask R-CNN,” in Proc. IEEE Int. Conf. Comput. Vis. , 2017, pp. 2980–2988 with faster inference.

30 August 2021 in IEEE Robotics and Automation Letters
Verified
Journal Article
Closed access
RCRuihang Chu
Ruihang Chu
YCYukang Chen
Yukang Chen
TKTao Kong
Tao Kong
LQLu Qi
Lu Qi
LLLei Li
Lei Li
Separating 3D point clouds into individual instances is an important task for 3D vision. It is challenging due to the unknown and varying number of instances in a scene. Existing deep learning based works focus on a two-step pipeline: first learn a feature embedding and then cluster the points. Such a two-step pipeline leads to disconnected intermediate objectives. In this paper, we propose an integrated reformulation of 3D instance segmentation as a per-point classification problem. We propose ICM-3D, a single-step method to segment 3D instances via instantiated categorization. The augmented category information is automatically constructed from 3D spatial positions. We conduct extensive experiments to verify the effectiveness of ICM-3D and show that it obtains inspiring performance across multiple frameworks, backbones and benchmarks.

12 April 2022 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Verified
Journal Article
Closed access
YCYukang Chen
Yukang Chen
PZPeizhen Zhang
Peizhen Zhang
TKTao Kong
Tao Kong
YLYanwei Li
Yanwei Li
XZXiangyu Zhang
Xiangyu Zhang
JJJiaya Jia
Jiaya Jia
Data augmentation is a critical technique in object detection, especially the augmentations targeting at scale invariance training (scale-aware augmentation). However, there has been little systematic investigation of how to design scale-aware data augmentation for object detection. We propose Scale-aware AutoAug to learn data augmentation policies for object detection. We define a new scale-aware search space, where both image- and instance-level augmentations are designed for maintaining scale robust feature learning. Upon this search space, we propose a new search metric, termed Pareto Scale Balance, to facilitate efficient augmentation policy search. In experiments, Scale-aware AutoAug yields significant and consistent improvement on various object detectors (e.g., RetinaNet, Faster R-CNN, Mask R-CNN, and FCOS), even compared with strong multi-scale training baselines. Our searched augmentation policies are generalized well to other datasets and instance-level tasks beyond object detection, e.g., instance segmentation. The search cost is much less than previous automated augmentation approaches for object detection, i.e., 8 GPUs across 2.5 days versus. 800 TPU-days. In addition, meaningful patterns can be summarized from our searched policies, which intuitively provide valuable knowledge for hand-crafted data augmentation design. Based on the searched scale-aware augmentation policies, we further introduce a dynamic training paradigm to adaptively determine specific augmentation policy usage during training. The dynamic paradigm consists of an heuristic manner for image-level augmentations and a differentiable copy-paste-based method for instance-level augmentations. The dynamic paradigm achieves further performance improvements to Scale-aware AutoAug without any additional burden on the long tailed LVIS benchmarks. We also demonstrate its ability to prevent over-fitting for large models, e.g., the Swin Transformer large model. Code and models are available at https://github.com/dvlab-research/SA-AutoAug .

11 January 2024 in Microbiology Spectrum
Verified
Journal Article
Open access
YWYue Wang
Yue Wang
YTYixin Tang
Yixin Tang
YCYukang Chen
Yukang Chen
GYGuangxi Yu
Guangxi Yu
XZXue Zhang
Xue Zhang
SGSong Gao
Song Gao
The monkeypox virus, which was declared as a Public Health Emergency of International Concern (PHEIC) by the World Health Organization (WHO), continues to cause infection cases worldwide. Given the risk of virus evolution, it is essential to identify monkeypox virus infection in a timely manner and isolate patients to prevent outbreaks so that the vulnerable population is protected and the emergence of dangerous monkeypox variants is restrained. In this study, a novel one-pot recombinase polymerase amplification-Clustered Regularly Interspaced Short Palindromic Repeats (RPA-CRISPR) assay for monkeypox based on a sucrose-aided multiphase aqueous system has been established with an ultra-high sensitivity of a single copy, which is 10 times higher than the existing RPA-CRISPR one-pot method. The entire reaction was completed within 60 min at 37°C. The detection method exhibited good specificity, effectively distinguishing monkeypox from other orthopoxviruses. The detection results could be observed by the naked eye under ultraviolet or blue light, making it highly suitable for home or limited healthcare settings. The assay has solved the incompatibility between the Cas12a cleavage reaction and the RPA reaction and shows good specificity, accuracy, and the rapidness and convenience important for point-of-care testing. It provides an effective tool for the early diagnosis of monkeypox and sets a technical example of using a multiphase aqueous system to establish one-pot RPA-CRISPR detection. IMPORTANCE: The monkeypox virus was declared as a Public Health Emergency of International Concern (PHEIC) by the World Health Organization (WHO) and continues to cause infection cases worldwide. Given the risk of virus evolution, it is essential to identify monkeypox virus infection in a timely manner to prevent outbreaks. This study establishes a novel one-pot recombinase polymerase amplification-Clustered Regularly Interspaced Short Palindromic Repeats (RPA-CRISPR) assay for monkeypox virus with an ultra-high sensitivity. The assay shows good specificity, accuracy, and the rapidness and convenience important for point-of-care testing. It provides an effective tool for the early diagnosis of monkeypox, which is useful for the prevention of an epidemic.

1 September 2022 in Journal of Physics: Conference Series
Verified
Conference Paper
Open access
BLBaoyu Li
Baoyu Li
NCNing Chen
Ning Chen
YCYukang Chen
Yukang Chen
XXXin Xie
Xin Xie
DFDapeng Fan
Dapeng Fan
According to the requirements of high resolution and compact space of a rotational-double-prism system, combined with its practical application characteristics of small load and low speed, an offset twist-roller frictional transmission mechanism is proposed. Compared with the clearance elimination gear transmission mechanism, it has the advantages of high output resolution, adjustable reduction ratio, simple structure and low cost. The influence of twist angle on the output characteristics of offset twist-roller frictional transmission mechanism is studied by experiment. The results show that the larger the twist angle, the smaller the driving torque of the driven wheel, the larger the transmission ratio of the mechanism, the larger the fluctuation range of output angular velocity and transmission ratio, and the more unstable the transmission. The transmission ratio of the mechanism in stable transmission is about 30, which is much higher than that in single-stage cylindrical gear drive. The output angular resolution of the mechanism is better than 0.005°, meeting the system requirements. Through the above analysis, it lays a foundation for the application of offset twist-roller frictional transmission mechanism in rotational double prisms system.