Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks181 from public data
- EnlightenGAN: Deep Light Enhancement Without Paired Supervision2,636
This paper proposes a highly effective unsupervised generative adversarial network, dubbed EnlightenGAN, that can be trained without low/normal-light image pairs, yet proves to generalize very well on various real-world test images.
- Understanding Convolution for Semantic Segmentation1,998
DUC is designed to generate pixel-level prediction, which is able to capture and decode more detailed information that is generally missing in bilinear upsampling, and a hybrid dilated convolution (HDC) framework in the encoding phase is proposed.
- 1,955
- 814
- Residual Local Feature Network for Efficient Super-Resolution282
The main idea is using three convolutional layers for residual local feature learning to simplify feature aggregation, which achieves a good trade-off between model performance and inference time and won the first place in the runtime track of the NTIRE 2022 efficient super-resolution challenge.
- 279
- 265
- 261
- Image Super-Resolution via Dual-State Recurrent Networks241
This paper explores new structures for SR based on this compact RNN view, leading to a dual-state design, the Dual-State Recurrent Network (DSRN), which performs favorably against state-of-the-art algorithms in terms of both memory consumption and predictive accuracy.
- Studying Very Low Resolution Recognition Using Deep Networks238
This work forms a dedicated deep learning method that allows for both the flexibility to combat the LR-HR domain mismatch, and the robustness to outliers, and achieves feature enhancement and recognition simultaneously.
- D3: Deep Dual-Domain Based Fast Restoration of JPEG-Compressed Images229
A Deep Dual-Domain (D3) based fast restoration model to remove artifacts of JPEG compressed images that leverages the large learning capacity of deep networks, as well as the problem-specific expertise that was hardly incorporated in the past design of deep architectures.
- Connecting Image Denoising and High-Level Vision Tasks via Deep Learning200
A convolutional neural network in which convolutions are conducted in various spatial resolutions via downsampling and upsampling operations in order to fuse and exploit contextual information on different scales and the benefit of exploiting image semantics simultaneously for image denoising and high-level vision tasks via deep learning is demonstrated.
- 199
- 167
- Machine Learning by Two-Dimensional Hierarchical Tensor Networks: A Quantum Information Theoretic Perspective on Deep Architectures152
This work trains two-dimensional hierarchical TNs to solve image recognition problems, using a training algorithm derived from the multi-scale entanglement renormalization ansatz, and introduces mathematical connections among quantum many-body physics, quantum information theory, and machine learning.
- Pyramid Attention Network for Image Restoration142
A novel Pyramid Attention module for image restoration, which captures long-range feature correspondences from a multi-scale feature pyramid, and is designed to be able to borrow clean signals from their “clean” correspondences at the coarser levels.
- Experiments on deep learning for speech denoising140
A very lightweight procedure that can predict clean speech spectra when presented with noisy speech inputs is proposed and it is concluded that it can perform better than some comparable single-channel approaches and that it is able to generalize well across various speakers, noise types and signal-to-noise ratios.
- 135
- Fast and Memory-Efficient Network Towards Efficient Image Super-Resolution132
A novel sequential attention branch is proposed, where every pixel is assigned an important factor according to local and global contexts, to enhance high-frequency details in a fast and memory-efficient network.
- 127
- 122
- 110
- 106
- 94
- NTIRE 2022 Challenge on Efficient Super-Resolution: Methods and Results94
The NTIRE 2022 challenge was to super-resolve an input image with a magnification factor of ×4 based on pairs of low and corresponding high resolution images and the aim was to design a network for single image super-resolution that achieved improvement of efficiency measured according to several metrics.
- Liveness of an extended S3PR88
This paper presents a new prevention method for deadlock issue arising in flexible manufacturing systems modeled with Petri nets based on a characteristic structure of WS^3PR, an extension of System of Simple Sequential Processes with Resources with Resources (S*3PR) with weighted arcs.
- Survey of Face Detection on Low-Quality Images84
The results demonstrate that both hand-crafted and deep-learning based face detectors are not robust enough for low-quality images, and inspires researchers to produce more robust design for face detection in the wild.
- CompFeat: Comprehensive Feature Aggregation for Video Instance Segmentation80
This work proposes a novel comprehensive feature aggregation approach (CompFeat) to refine features at both frame-level and object-level with temporal and spatial context information to eliminate ambiguities introduced by only using single-frame features.
- 78
- Progressive Temporal Feature Alignment Network for Video Inpainting72
This work proposes a ‘Progressive Temporal Feature Alignment Network’, which progressively enriches features extracted from the current frame with the feature warped from neighbouring frames using optical flow, greatly improving visual quality and temporal consistency of the inpainted videos.
- 69
- On Randomized Broadcasting and Gossiping in Radio Networks69
An O(N log3N)- time randomized algorithm for gossiping in radio networks with unknown topology is given, an improvement over the fastest previously known algorithm that works in time O( N log4 N).
- ShadowFormer: Global Context Helps Image Shadow Removal67
A Retinex-based shadow model is proposed, from which a novel transformer-based network is derived, dubbed ShandowFormer, to exploit non-shadow regions to help shadow region restoration and achieves state-of-the-art performance by using up to 150X fewer model parameters.
- 61
- Property-Preserving Data Reconstruction57
A randomized filter that can answer any query in O( log 2nlog log’n) time while modifying the function f at only O(εn) places is presented, and the amortized time over n function evaluations is O(log”n).
- Adversarial Open Domain Adaptation for Sketch-to-Photo Synthesis54
This paper proposes a simple yet effective open-domain sampling and optimization strategy to "fool" the generator into treating fake sketches as real ones, and shows impressive results in synthesizing realistic color, texture, and maintaining the geometric composition for various categories of open- domain sketches.
- Generative tensor network classification model for supervised machine learning53
A generative TN classification (GTNC) approach for supervised learning that is more efficient than the existing TN models that are in general discriminative and an adaptive and universal model of excellent performance.
- 47
- Scale-Wise Convolution for Image Restoration46
It is shown that properly modeling scale-invariance into neural networks can bring significant benefits to image restoration performance and the proposed network with scale-wise convolution achieves superior performance in multiple image restoration tasks including image super-resolution, image denoising and image compression artifacts removal.
- 45
- 41
- 41
- 41
- 40
- Deeply Improved Sparse Coding for Image Super-Resolution.39
This paper shows that a sparse coding model particularly designed for super-resolution can be incarnated as a neural network, and trained in a cascaded structure from end to end, and leads to much more efficient and effective training, as well as a reduced model size.
- Learning Object Detectors from Scratch with Gated Recurrent Feature Pyramids.38
This study proposes a recurrent feature-pyramid structure to squeeze rich spatial and semantic features into a single prediction layer that further reduces the number of parameters to learn, and is the best performed model of learning object detection from scratch.
- Learning Simple Thresholded Features With Sparse Support Recovery35
TheDLTF model, which learns an optimized dictionary for applying the thresholded feature, is formulated and evaluated on a vast range of synthetic and real-data tasks, where DLTF demonstrates remarkable efficiency, effectiveness, and robustness in all experiments.
- Learning a Mixture of Deep Networks for Single Image Super-Resolution33
This paper proposes the method of learning a mixture of SR inference modules in a unified framework to tackle single image super-resolution problem, and achieves state-of-the-arts restoration results on a wide range of images consistently while allowing more flexible design choices.
- Sparse Coding and its Applications in Computer Vision33
This book provides a broader introduction to the theories and applications of sparse coding techniques in computer vision research, and introduces sparse coding in the context of representation learning, illustrates the fundamental concepts, and summarizes the most active research directions.
- Unsupervised Low-light Image Enhancement with Decoupled Networks32
This paper proposes a two-stage, fully unsupervised model to handle both illumination enhancement and noise suppression of real-world low-light images, and shows that the proposed method outperforms the state-of-the-art unsuper supervised methods concerning both illumination Enhancement and noise reduction.
- Extended elementary siphon-based deadlock prevention policy for a class of generalised Petri nets32
This paper develops a novel deadlock prevention policy for a class of generalised Petri nets, namely S4PR that can well model flexible manufacturing systems (FMS), and proposes the concept of augmented siphons, a concept of max′-controlled siphons that can relax the siphon controllability condition.
- 32
- Optimal enforcement of liveness to flexible manufacturing systems modeled with Petri nets via transition-based controllers30
A transition-based control technique to prevent deadlocks for flexible manufacturing systems that can be modeled with a generalized class of Petri nets using the structural properties of the Petri net model to avoid the computation of its reachability graph which in general leads to the state explosion problem.
- Robust emotion recognition from low quality and low bit rate video: A deep learning approach30
A novel framework to achieve robust emotion recognition from low bit rate video with significantly improved stand-alone recognition performance, as well as rate-distortion (R-D) performance, than either directly recognizing from LR frames, or separating SR and recognition.
- 27
- Approximate range searching in higher dimension27
An approximate answer to a range query lowers the storage requirement from exponential to polynomial, and generalize the solution to polytope/ball range searching.
- Lower bounds for intersection searching and fractional cascading in higher dimension27
It is proved that the answer to all of these questions is no, and near-optimal solutions to these classical problems are derived.
- 25
- 25
- Natural Melanin-Based Nanoparticles With Combined Chemo/Photothermal/Photodynamic Effect Induce Immunogenic Cell Death (ICD) on Tumor25
It was suggested that PDA-DOX NPs with NIR irradiation could serve as a promising agent for tumor therapy.
- Unsupervised Real-world Low-light Image Enhancement with Decoupled Networks.25
This paper proposes to learn a two-stage GAN-based framework to enhance the real-world low-light images in a fully unsupervised fashion and shows that this method outperforms the state-of-the-art unsuper supervised image enhancement methods in terms of both illumination enhancement and noise reduction.
- 25
- A parameterized liveness and ratio-enforcing supervisor for a class of generalized Petri nets22
A new kind of supervisors is synthesized on the ground of the generic nature of the intrinsically live structure that can achieve the purposes of both liveness-enforcement and resource usage ratio-enforcement of the system under consideration.
- The Hydrogen Energy Infrastructure Location Selection Model: A Hybrid Fuzzy Decision-Making Approach21
- 20
- AdaFormer: Efficient Transformer with Adaptive Token Sparsification for Image Super-resolution18
A texture-relevant sparse attention block with parallel global and local branches is introduced, aiming to integrate informative tokens from the global view instead of only in fixed local windows, and an early-exit strategy is designed to progressively halt tokens according to the token importance.
- Dynamic Proposals for Efficient Object Detection18
This paper designs a module to make a single query-based model to be able to inference with different numbers of proposals, and extends it to a dynamic model to choose the number of proposals according to the input image, greatly reducing computational costs.
- Learning audio sequence representations for acoustic event classification17
An unsupervised learning framework to learn a vector representation of an audio sequence for AEC is proposed, which can not only deal with the problem of arbitrary-lengths of audio streams, but also learn the salient information of the sequence.
- Boosting Lightweight Single Image Super-resolution via Joint-distillation16
A novel joint-distillation (JDSR) framework to boost the representation of various off-the-shelf lightweight SR models by a large margin, with the same model architecture and inference expense.
- 15
- 15
- 14
- 14
- 13
- 13
- 13
- A Minimal Supervisory Structure to Optimally Enforce Liveness on Petri Net Models for Flexible Manufacturing Systems13
A new method of computing a minimal supervisory structure that optimally enforces liveness on the Petri net models for flexible manufacturing systems (FMSs) is presented, which provides an approach to design control places for each computed GMEC without solving integer linear programming problems, which greatly reduces the computational costs.
- Visual Recognition in Very Low-Quality Settings: Delving Into the Power of Pre-Training11
This paper presents a degradation-robust pre-training approach on improving deep learning models towards this direction, and extensive experiments on different datasets validate the effectiveness of the proposed method.
- 10
- 9
- 9
- 9
- Secure WDM-PON based on chaos synchronization and subcarrier modulation multiplexing9
A novel secure passive optical network (PON) configuration jointly based on the chaos synchronization of wavelength division multiplexed (WDM) chaotic signals and the subcarrier modulation multiplexing is introduced and can improve the bandwidth utilization efficiency and obviously enhance the information security at the physical layer of the WDM-PON.
- A fast algorithm for denoising magnitude diffusion‐weighted images with rank and edge constraints9
To accelerate denoising of magnitude diffusion‐weighted images subject to joint rank and edge constraints, a new approach is proposed to combine laser-spot assisted and 3D image analysis.
- Joint frequency, 2-D DOA, and polarization estimation using parallel factor analysis9
This paper proposes a new algorithm for joint frequency, two-dimensional directions-of-arrival (DOA), and polarization estimation using parallel factor (PARAFAC) analysis model and cumulant, which alleviates the aperture loss, and avoids pairing parameter.
- Deadlock Control for a Class of Generalized Petri Nets Based on Proper Resource Allocation8
A deadlock prevention policy is developed for a class of generalized Petri nets, namely G‐systems, which can model flexible manufacturing systems with machining, assembly, and disassembly operations and can usually lead to a suboptimal supervisor with high computational efficiency.
- 8
- 7
- Diameter Prediction of Silicon Ingots in the Czochralski Process Based on a Hybrid Deep Learning Model7
A novel hybrid deep learning model is presented, which combines the deep belief network (DBN), support vector regression (SVR), and the ant lion optimizer (ALO) to address the diameter prediction of silicon ingots in the Czochralski process.
- 7
- 7
- 7
- Adjustable Memory-efficient Image Super-resolution via Individual Kernel Sparsity6
This work essentially provides a dynamic parameter reassignment scheme with a given resource budget for an off-the-shelf SR model and demonstrates that IKS imparts considerable sparsity with negligible effect on SR quality.
- A Blockchain-Based Security Model for IoT Systems6
This paper proposes a smart contract-based security model for IoT systems based on the super ledger Fabric blockchain platform having decentralised, tamper-proof, and programmable features, and gains a trustworthy environment for enhancing the security of the whole IoT system.
- Connecting Low-Level Image Processing and High-Level Vision via Deep Learning6
The goal is to design deep learning mechanisms that can efficiently and effectively learn features from low-level image processing and use them to improve the performance of high-level vision tasks.
- 6
- Data-driven clustering learning modeling for semiconductor silicon single crystal growth process based on fuzzy C-means and DBN5
This study introduces a novel approach for modeling clustering based on data that integrate fuzzy C-means (FCM) and deep belief network (DBN) techniques and suggests that the method put forth exhibits superior precision and enhanced resilience in predicting crystal diameter and melt temperature compared to alternative modeling approaches.
- 5
- 5
- 4
- 4
- 4
- 4
- 4
- An Efficient Energy-Saving Scheme Based on Grouping of ONU for Optical Access Network Using Electronic Switch4
It is demonstrated that obvious energy-saving effects can be achieved, and the energy- saving efficiency can be further improved by expanding the capacity of the links between RN and M-ONUs and properly selecting the scale of the ONU group.
- On Intrinsically Live Structure of a Class of Generalized Petri Nets Modeling FMS4
The numerical relationship between initial markings of resource places and weights of arcs is investigated to yield an intrinsically live structure-based liveness-enforcing method that requires no external monitors.
- 4
- Network meta-analysis of different electrical stimulation therapies for lower limb functional rehabilitation in stroke patients3
Different electrical stimulation modalities provide domain-specific benefits in post-stroke lower limb rehabilitation, and Tailored selection of stimulation techniques may optimize functional recovery.
- 3
- 3
- Elliptic Arc Fitting for the Shape Control System of Silicon Single Crystal Growth3
A two-side (TS) constraint method to fit the elliptic arc parameters effectively and two weighted functions, which give the data out of TS constraint a small weight coefficient to resist the large level noise disturbing, are brought into the TS constraint optimization problem tactfully.
- 3
- 3
- Robot Simultaneous Localization and Mapping Based on Path Planning in Dynamic Environments3
The SLAM problem in dynamic environments with random objects characterized by obstacle is solved by the proposed algorithm and relevant performance indicators prove that the estimation of the algorithm is consistent.
- 3
- 2
- 2
- 2
- A Redundancy Based Unsupervised Feature Selection Method for High-Dimensional Data2
A redundancy based unsupervised feature selection method for high-dimensional data called as RUFS that improves the mean classification of 11 real datasets by 8.1 percent at least.
- 2
- A Survey of Multimodal Fusion for Identity Verification2
The knowledge of multimodal fusion is introduced, some commonly used multi-modal fusion techniques in the field of identity verification are listed, and some prospects are put forward for the future multimodals fusion identity verification.
- Signal Processing2
The experimental results demonstrate that the DUpsampling strategy achieves state-of-the-art performance with competing methods, and can achieve the segmentation speed as 90.84FPS and the segmentation accuracy as 90.770 % mIoU.
- Single Image Super-Resolution: From Sparse Coding to Deep Learning2
It is demonstrated that a sparse coding model particularly designed for super-resolution can be incarnated as a neural network with the merit of end-to-end optimization over training data and significantly outperforms existing state-of-the-art methods for various scaling factors both quantitatively and qualitatively.
- 2
- Epitomic Image Super-Resolution2
Instead of local nearest neighbor patch matching used in most existing internal SR methods, ESR employs epitomic patch matching that features robustness to noise, and both local and non-local patch matching.
- Surface position detection method of silicon melt in CZ furnace2
A triangulation method based on the line laser and area image sensor for silicon melt surface position detection in the Czochralski mono-crystalline silicon crystal growth system and has a good real-time performance and robustness.
- 2
- 2
- 2
- 1
- 1
- 1
- 1
- A Practical Routing Protocol Based on WiFi for Resources-constrained Opportunistic Networks1
An opportunistic networks routing protocol based on WiFi (ONRWF) which aims to make up for the lack of bandwidth and bufier space and experimental results demonstrate signiflcant performance beneflts in terms of message delivery.
- 1
- 1
- Theory and methodology of object-oriented formal modelling1
A strategy to improve the correctness of formal models is proposed by establishing, validating and verifying the function model and data structure separately, and fusing them finally, as the following steps.
- 1
- 1
- 1
- –
- –
- Research on TOD Development and Location Selection among All Urbanized Areas – A case study in China–
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Robust Outdoor Vehicle Visual Tracking Based on k-Sparse Stacked Denoising Auto-Encoder–
K-sparse constraint is added to the encoder part of stacked auto-encoder network to learn more invariant feature of object appearance, and a stacked k-s parse-autoencoder–based robust outdoor vehicle tracking method under particle filter inference is further proposed to solve the problem of appearance variance during the tracking.
- –
- –
- –
- –
- –
- –
- –
- Diagnosis of interstitial fibrosis of renal allograft by non-linear optical imaging technology–
Non-linear optical technology can clearly and early display renal allograft interstitial fibrosis and tissue structure and SHG/TPEF is a better method than traditional staining method for accessing renal allogsraft interinterstitial fibrosis.
- –
- Extension of Elementary Siphons in a Class of Generalized Petri Nets–
Through fully investigating the structure of the net, especially weight information, the set of elementary siphons obtained by the proposed method is more compact and well suits for WS3PR.
- –
- CMPR: COST-AWARE MULTIPATH ROUTING ALGORITHM FOR MOBILE WIRELESS SENSOR NETWORKS–
Simulation results show that the proposed cost-aware multipath routing algorithm evidently achieves better performance in terms of the balance of node energy consumption, the ratio of received packets, the end-to-end transmission delay and the delay jitter.
- –
- A system for developing a series of interactive tests of vocal production requiring on-line audiovisual recording–
AIRS-TEST administers a sequence of interactive tests and organizes the results for analysis, and an authority module is associated with collected data to control user's right of retrieval, considering both confidentiality and collaborative sharing.
- –
- Passive Source Localization of Sensor Arrays–
Several high-resolution methods are presented to overcome difficulties in passive source localization and for joint azimuth and elevation direction finding, especially when elevation angles are between 70 and 90 degrees.
- –
- –
- –
- –
- –
- Prevention Analysis on Malicious Code of Mobile Phone–
The characteristic prototypes and attacking approaches of malicious codes are concluded so as to analyze the principle of malicious code and forecast its developmental trend.
- –
- –
- –
- –
- Application MMX in Video Compression–
The results show that the optimization on instruction level can heighten video coder's running speed distinctly and improve the quality of video compression H. 263-based films.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.