A5101773973- Advanced Image and Video Retrieval Techniques
- Advanced Vision and Imaging
- Visual Attention and Saliency Detection
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Advanced Image Processing Techniques
- Multimodal Machine Learning Applications
- Image Enhancement Techniques
- Advanced Wireless Network Optimization
- Advanced MIMO Systems Optimization
- Human Pose and Action Recognition
- Generative Adversarial Networks and Image Synthesis
- Anomaly Detection Techniques and Applications
- VLSI and FPGA Design Techniques
- Video Analysis and Summarization
- Wireless Communication Networks Research
- Cooperative Communication and Network Coding
- Domain Adaptation and Few-Shot Learning
- UAV Applications and Optimization
- Embedded Systems Design Techniques
- Underwater Vehicles and Communication Systems
- Medical Image Segmentation Techniques
- Low-power high-performance VLSI design
- Interconnection Networks and Systems
- Image Processing Techniques and Applications
Wuhan University of Technology
2009-2025
Universitas Nurtanio
2024
University of Essex
2024
Adobe Systems (United States)
2018-2023
Agency for Science, Technology and Research
2023
Shanghai Jiao Tong University
2011-2020
Beijing Institute of Technology
2020
Yonsei University
2019
Amazon (United States)
2019
Seoul National University
2019
We propose a novel solution for semi-supervised video object segmentation. By the nature of problem, available cues (e.g. frame(s) with masks) become richer intermediate predictions. However, existing methods are unable to fully exploit this rich source information. resolve issue by leveraging memory networks and learn read relevant information from all sources. In our framework, past frames masks form an external memory, current frame as query is segmented using mask in memory....
Image matting is a fundamental computer vision problem and has many applications. Previous algorithms have poor performance when an image similar foreground background colors or complicated textures. The main reasons are prior methods 1) only use low-level features 2) lack high-level context. In this paper, we propose novel deep learning based algorithm that can tackle both these problems. Our model two parts. first part convolutional encoder-decoder network takes the corresponding trimap as...
In this paper we present a new computer vision task, named video instance segmentation. The goal of task is simultaneous detection, segmentation and tracking instances in videos. words, it the first time that image problem extended to domain. To facilitate research on propose large-scale benchmark called YouTube-VIS, which consists 2,883 high-resolution YouTube videos, 40-category label set 131k high-quality masks. addition, novel algorithm MaskTrack R-CNN for task. Our method introduces...
Learning long-term spatial-temporal features are critical for many video analysis tasks. However, existing segmentation methods predominantly rely on static image techniques, and capturing temporal dependency have to depend pretrained optical flow models, leading suboptimal solutions the problem. End-to-end sequential learning explore spatialtemporal is largely limited by scale of available datasets, i.e., even largest dataset only contains 90 short clips. To solve this problem, we build a...
Artistic text style transfer is the task of migrating from a source image to target create artistic typography. Recent methods have considered texture control enhance usability. However, controlling stylistic degree in terms shape deformation remains an important open challenge. In this paper, we present first network that allows for real-time crucial glyph through adjustable parameter. Our key contribution novel bidirectional matching framework establish effective glyph-style mapping at...
We propose Mask Guided (MG) Matting, a robust matting framework that takes general coarse mask as guidance. MG Matting leverages network (PRN) design which encourages the model to provide self-guidance progressively refine uncertain regions through decoding process. A series of guidance perturbation operations are also introduced in training further enhance its robustness external show PRN can generalize unseen types masks such trimap and low-quality alpha matte, making it suitable for...
We propose a novel video inpainting algorithm that simultaneously hallucinates missing appearance and motion (optical flow) information, building upon the recent 'Deep Image Prior' (DIP) exploits convolutional network architectures to enforce plausible texture in static images. In extending DIP we make two important contributions. First, show coherent is possible without priori training. take generative approach based on internal (within-video) learning reliance an external corpus of visual...
We present a deep learning method for the interactive video object segmentation. Our is built upon two core operations, interaction and propagation, each operation conducted by Convolutional Neural Networks. The networks are connected both internally externally so that trained jointly interact with other to solve complex segmentation problem. propose new multi-round training scheme can learn how understand user's intention update incorrect estimations during training. At testing time, our...
Image matting is a key technique for image and video editing composition. Conventionally, deep learning approaches take the whole input an associated trimap to infer alpha matte using convolutional neural networks. Such set state-of-the-arts in matting; however, they may fail real-world applications due hardware limitations, since images are mostly of very high resolution. In this paper, we propose HDMatt, first based approach high-resolution inputs. More concretely, HDMatt runs patch-based...
We propose a novel and unified solution for user-guided video object segmentation tasks. In this work, we consider two scenarios of segmentation: semi-supervised interactive segmentation. Due to the nature problem, available cues - frame(s) with masks (or scribbles) become richer intermediate predictions additional user inputs). However, existing methods make it impossible fully exploit rich source information. resolve issue by leveraging memory networks learning read relevant information...
Interactive object cutout tools are the cornerstone of image editing workflow. Recent deep-learning based interactive segmentation algorithms have made significant progress in handling complex images and rough binary selections can typically be obtained with just a few clicks. Yet, deep learning techniques tend to plateau once this selection has been reached. In work, we interpret as inability current sufficiently leverage each user interaction also limitations training/testing datasets. We...
In this study, we present a novel design methodology for unit cells in chessboard metasurfaces with the aim of reducing radar cross-section (RCS) linearly polarized waves. The employs rotational symmetry and incorporates ten continuous parameters to define metasurface units, enabling creation flexible 2D structures. geometrical two units are then optimized using simulated annealing (SA) algorithm achieve low RCS metasurface. Following optimization, properties were experimentally verified....
In this paper, an occlusion handling algorithm is presented for motion-compensated frame interpolation (MCFI). The proposed first estimates the unidirectional motion vector fields (MVFs) between two reference frames in both forward and backward directions. Then, a multiframe-based detection method to identify uncovered, covered, ambiguity regions within intermediate plane. Based on results, bidirectional vectors of are assigned accurately from MVFs. For compensation, conventional overlapped...
Learning long-term spatial-temporal features are critical for many video analysis tasks. However, existing segmentation methods predominantly rely on static image techniques, and capturing temporal dependency have to depend pretrained optical flow models, leading suboptimal solutions the problem. End-to-end sequential learning explore is largely limited by scale of available datasets, i.e., even largest dataset only contains 90 short clips. To solve this problem, we build a new large-scale...
Designing the structure of neural networks is considered one most challenging tasks in deep learning, especially when there few prior knowledge about task domain. In this paper, we propose an Ecologically-Inspired GENetic (EIGEN) approach that uses concept succession, extinction, mimicry, and gene duplication to search network from scratch with poorly initialized simple constraints forced during evolution, as assume no Specifically, first use primary succession rapidly evolve a population...
Video enhancement plays an important role in various video applications. In this paper, we propose a new intra-and-inter-constraint-based approach aiming to 1) achieve high intra-frame quality of the entire picture where multiple region-of-interests (ROIs) can be adaptively and simultaneously enhanced, 2) guarantee inter-frame consistencies among frames. We first analyze features from different ROIs create piecewise tone mapping curve for frame such that enhanced. further introduce...
Video instance segmentation is a challenging task that extends image to the video domain. Existing methods either rely only on single-frame information for detection and subproblems or handle tracking as separate post-processing step, which limit their capability fully leverage share useful spatial-temporal all subproblems. In this paper, we propose novel graph-neural-network (GNN) based method aforementioned limitation. Specifically, graph nodes representing features are used while edges...
Videos contain various types and strengths of motions that may look unnaturally discontinuous in time when the recorded frame rate is low. This paper reviews first AIM challenge on video temporal super-resolution (frame interpolation) with a focus proposed solutions results. From low-frame-rate (15 fps) sequences, participants are asked to submit higher-frame-rate (60 sequences by estimating temporally intermediate frames. We employ REDS_VTSR dataset derived from diverse videos captured...
Unmanned Surface Vehicles (USVs) in inland waterways have drawn increasing attention for their excellent capability to serve maritime time-consuming missions such as autonomous navigation and intelligent monitoring. However, USVs struggle accomplish emerging computation-intensive tasks (e.g., sensor, telemetry, etc) timely due the limited on-board resources. This paper proposes a novel reconfigurable surface (RIS)-assisted unmanned aerial vehicle (UAV) multi-access edge computing (MEC)...