A5101631518- Advanced Image Processing Techniques
- Multimodal Machine Learning Applications
- Advanced Vision and Imaging
- Image Enhancement Techniques
- Face recognition and analysis
- Generative Adversarial Networks and Image Synthesis
- Advanced Image and Video Retrieval Techniques
- Augmented Reality Applications
- Legionella and Acanthamoeba research
- Tactile and Sensory Interactions
- AI in cancer detection
- Photodynamic Therapy Research Studies
- Robotics and Sensor-Based Localization
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Advanced Memory and Neural Computing
- Domain Adaptation and Few-Shot Learning
- Image and Signal Denoising Methods
- Corporate Finance and Governance
- Ferroelectric and Negative Capacitance Devices
- Financial Markets and Investment Strategies
- Image Processing Techniques and Applications
- Brain Tumor Detection and Classification
- Advancements in Photolithography Techniques
- Interactive and Immersive Displays
Peng Cheng Laboratory
2024
UC San Diego Health System
2023
Tianjin Stomatological Hospital
2023
University of Science and Technology of China
2021-2022
University of Science and Technology Chittagong
2022
North Carolina State University
2017
Deep learning-based methods have achieved remarkable performance for image dehazing. However, previous studies are mostly focused on training models with synthetic hazy images, which incurs drop when the used real-world images. We propose a Principled Synthetic-to-real Dehazing (PSD) framework to improve generalization of Starting from dehazing model backbone that is pre-trained data, PSD exploits real images fine-tune in an unsupervised fashion. For fine-tuning, we leverage several...
Videos typically record the streaming and continuous visual data as discrete consecutive frames. Since storage cost is expensive for videos of high fidelity, most them are stored in a relatively low resolution frame rate. Recent works Space-Time Video Super-Resolution (STVSR) developed to incorporate temporal interpolation spatial super-resolution unified framework. However, only support fixed up-sampling scale, which limits their flexibility applications. In this work, instead following...
Abstract Dental caries is a common disease caused by plaque biofilms, which are important pathogenic factors in many diseases. When hosts overexposed to dietary sugars, pathogens such as Streptococcus mutans (S. mutans) and other cariogenic bacteria, metabolically assemble an extracellular matrix rich exopolysaccharides form disease‐causing biofilm, the microenvironment characterized regional hypoxia, low pH, nutritional deprivation. Current antimicrobials with inadequate penetration lack of...
We present a deep learning approach for high resolution face completion with multiple controllable attributes (e.g., male and smiling) under arbitrary masks. Face entails understanding both structural meaningfulness appearance consistency locally globally to fill in "holes" whose content do not appear elsewhere an input image. It is challenging task the difficulty level increasing significantly respect resolution, complexity of filled-in fragments. Our system addresses challenges by fully...
Vision Language Models (VLMs), which extend Large (LLM) by incorporating visual understanding capability, have demonstrated significant advancements in addressing open-ended question-answering (VQA) tasks. However, these models cannot accurately interpret images infused with text, a common occurrence real-world scenarios. Standard procedures for extracting information from often involve learning fixed set of query embeddings. These embeddings are designed to encapsulate image contexts and...
Low-light images captured in the real world are inevitably corrupted by sensor noise. Such noise is spatially variant and highly dependent on underlying pixel intensity, deviating from oversimplified assumptions conventional denoising. Existing light enhancement methods either overlook important impact of real-world during enhancement, or treat removal as a separate pre- post-processing step. We present \underline{C}oordinated \underline{E}nhancement for \underline{R}eal-world...
Incorporating human feedback has been shown to be crucial align text generated by large language models preferences. We hypothesize that state-of-the-art instructional image editing models, where outputs are based on an input and instruction, could similarly benefit from feedback, as their may not adhere the correct instructions preferences of users. In this paper, we present a novel framework harness for visual (HIVE). Specifically, collect edited images learn reward function capture...
Performing holistic 3D scene understanding from a single-view observation, involving generating instance shapes and segmentation, is long-standing challenge. Prevailing works either focus only on geometry or model the task in two folds by separate modules, whose results are merged later to form final prediction. Inspired recent advances 2D vision that unify image segmentation detection Transformer-based models, we present Uni-3D, parsing/reconstruction system for single RGB image. Uni-3D...
This paper describes a prototype tangible six degree of freedom (6 DoF) input device that is inexpensive and intuitive to use: cube with colored corners specific shapes, tracked by single camera, pose estimated in real time. A tracking automatic color adjustment system are designed so the can work robustly noisy surroundings invariant changes lighting background noise. evaluation shows good performance for both refresh (above 60 FPS on average) accuracy estimation (average angular error...