A5102776532- Image Enhancement Techniques
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Quantum Dots Synthesis And Properties
- Chalcogenide Semiconductor Thin Films
- Image and Signal Denoising Methods
- Advanced Image Fusion Techniques
- Nanowire Synthesis and Applications
- ZnO doping and properties
- Computer Graphics and Visualization Techniques
- Fire Detection and Safety Systems
- Advanced Neural Network Applications
- Video Surveillance and Tracking Methods
- Gold and Silver Nanoparticles Synthesis and Applications
- Air Quality Monitoring and Forecasting
- Layered Double Hydroxides Synthesis and Applications
- Anodic Oxide Films and Nanostructures
- Advanced Technologies in Various Fields
- TiO2 Photocatalysis and Solar Cells
- Analytical Chemistry and Sensors
- Image Processing Techniques and Applications
- Magnesium Oxide Properties and Applications
- Rough Sets and Fuzzy Logic
- Neural Networks and Reservoir Computing
- Fault Detection and Control Systems
Nantong University
2025
University of Science and Technology of China
1999-2024
Shanghai Artificial Intelligence Laboratory
2023
Beijing Academy of Artificial Intelligence
2023
Hefei University
2023
Deep convolutional neural networks (CNNs) have become dominant in the single image de-raining area. However, most deep CNNs-based methods are designed by stacking vanilla layers, which can only be used to model local relations. Therefore, long-range contextual information is rarely considered for this specific task. To address above problem, we propose a simple yet effective dual graph network (GCN) rain removal. Specifically, design two graphs perform global relational modeling and...
Image restoration under multiple adverse weather conditions aims to remove weather-related artifacts by using a single set of network parameters. In this paper, we find that image degradations different contain general characteristics as well their specific characteristics. Inspired observation, design an efficient unified framework with two-stage training strategy explore the weather-general and weather-specific features. The first stage learn features taking images various inputs...
Shadow removal, which aims to restore the background in shadow regions, is challenging due its highly ill-posed nature. Most existing deep learning-based methods individually remove by only considering content of matched paired images, barely taking into account auxiliary supervision generation removal procedure. In this work, we argue that and are interrelated could provide useful informative for each other. Specifically, propose a new Bijective Mapping Network (BMNet), couples learning...
This paper reviews the NTIRE 2023 challenge on efficient single-image super-resolution with a focus proposed solutions and results. The aim of this is to devise network that reduces one or several aspects such as runtime, parameters, FLOPs, activations, memory footprint, depth RFDN while at least maintaining PSNR 29.00dB DIV2K validation datasets. had 272 registered participants, 35 teams made valid submissions. They gauge state-of-the-art for super-resolution.
Image deraining is a challenging task since rain streaks have the characteristics of spatially long structure and complex diversity. Existing deep learning-based methods mainly construct networks by stacking vanilla convolutional layers with local relations, can only handle single dataset due to catastrophic forgetting, resulting in limited performance insufficient adaptability. To address these issues, we propose new image framework effectively explore nonlocal similarity, continuously...
In this paper, we present a novel shadow detection framework by investigating the mutual complementary mechanisms contained in specific task. Our method is based on key observation: single image, regions and non-shadow counterparts are to each other nature, thus better estimation one side leads an improved other, vice versa. Motivated observation, first leverage two parallel interactive branches jointly produce masks. The interaction between retain deactivated intermediate features of branch...
In this work, we introduce a dual transformation network for single image contrast enhancement, which usually aims to improve global and enrich local details. To end, propose two parallel branches respectively handle the goals by learning different kinds of transformations. Specifically, one branch construct curve contrast, while other directly predicts pixel offsets addition, further design differentiable histogram loss provide supervised information related contrast. way, training can be...
Reflection from glasses is ubiquitous in daily life, but it usually undesirable photographs. To remove these unwanted noises, existing methods utilize either correlative auxiliary information or handcrafted priors to constrain this ill-posed problem. However, due their limited capability describe the properties of reflections, are unable handle strong and complex reflection scenes. In article, we propose a hue guidance network (HGNet) with two branches for single image removal (SIRR) by...
Abstract Nanowires and coral-shaped nanostructures of Ag were prepared by an ultraviolet photo-reduction technique at room temperature.
Albeit existing deep learning-based image de-raining methods have achieved promising results, most of them only extract single scale features, and neglect the fact that similar rain streaks appear repeatedly across different scales. Therefore, this paper aims to explore cross-scale cues in a multi-scale fashion. Specifically, we first introduce an adaptive-kernel pyramid provide effective information. Then, design two similarity attention blocks (CSSABs) search spatial channel relationships...
Abstract Polyacrylamide (PAM)-MS (M = Cd, Zn, Pb) inorganic-polymer nanocomposites with homogeneously dispersed semiconductor nanoparticles of narrow size distribution in the polymer matrices were prepared by a novel situ ultraviolet irradiation polymerization–photolysis (UIPP) technique at room temperature.
The increasing demand for computational photography and imaging on mobile platforms has led to the widespread development integration of advanced image sensors with novel algorithms in camera systems. However, scarcity high-quality data research rare opportunity in-depth exchange views from industry academia constrain intelligent (MIPI). Building achievements previous MIPI Workshops held at ECCV 2022 CVPR 2023, we introduce our third challenge including three tracks focusing algorithms. In...
Hybrid Event-Based Vision Sensor (HybridEVS) is a novel sensor integrating traditional frame-based and event-based sensors, offering substantial benefits for applications requiring low-light, high dynamic range, low-latency environments, such as smartphones wearable devices. Despite its potential, the lack of Image signal processing (ISP) pipeline specifically designed HybridEVS poses significant challenge. To address this challenge, in study, we propose coarse-to-fine framework named...