A5088829322- Advanced Image Processing Techniques
- Image and Signal Denoising Methods
- Advanced Vision and Imaging
- Image Enhancement Techniques
- Image Processing Techniques and Applications
- Generative Adversarial Networks and Image Synthesis
- Digital Media Forensic Detection
- Face recognition and analysis
- Image and Video Quality Assessment
- Color Science and Applications
- Medical Imaging and Analysis
- Digital Imaging for Blood Diseases
- Advanced Image and Video Retrieval Techniques
- Optical measurement and interference techniques
- Malaria Research and Control
- Mosquito-borne diseases and control
- 3D Surveying and Cultural Heritage
- Fire Detection and Safety Systems
- Video Analysis and Summarization
- COVID-19 diagnosis using AI
- 3D Shape Modeling and Analysis
- Multimodal Machine Learning Applications
- Integrated Circuits and Semiconductor Failure Analysis
- Optical Coherence Tomography Applications
- Human Pose and Action Recognition
Johns Hopkins University
2023-2024
Indian Institute of Technology Madras
2019-2023
Seoul National University
2021
Huawei Technologies (Sweden)
2019
ETH Zurich
2019
Polytechnic University of Timişoara
2019
This paper tackles the problem of motion deblurring dynamic scenes. Although end-to-end fully convolutional designs have recently advanced state-of-the-art in non-uniform deblurring, their performance-complexity trade-off is still sub-optimal. Existing approaches achieve a large receptive field by increasing number generic convolution layers and kernel-size, but this comesat expense increase model size inference speed. In work, we propose an efficient pixel adaptive feature attentive design...
This paper reviews the AIM 2019 challenge on real world super-resolution. It focuses participating methods and final results. The addresses setting, where paired true high low-resolution images are unavailable. For training, only one set of source input is therefore provided in challenge. In Track 1: Source Domain aim to super-resolve such while preserving low level image characteristics domain. 2: Target a high-quality also for that defines output domain desired quality super-resolved...
We present a general learning-based solution for restoring images suffering from spatially-varying degradations. Prior approaches are typically degradation-specific and employ the same processing across different pixels within. However, we hypothesize that such spatially rigid is suboptimal simultaneously degraded as well reconstructing clean regions of image. To overcome this limitation, propose SPAIR, network design harnesses distortion-localization information dynamically adjusts...
Motion blur is a common photography artifact in dynamic environments that typically comes jointly with the other types of degradation. This paper reviews NTIRE 2021 Challenge on Image Deblurring. In this challenge report, we describe specifics and evaluation results from 2 competition tracks proposed solutions. While both aim to recover high-quality clean image blurry image, different artifacts are involved. track 1, images low resolution while compressed JPEG format. each competition, there...
This paper reviews the second NTIRE challenge on image dehazing (restoration of rich details in hazy image) with focus proposed solutions and results. The training data consists from 55 images (with dense haze generated an indoor or outdoor environment) their corresponding ground truth (haze-free) same scene. has been produced using a professional haze/fog generator that imitates real conditions scenes. evaluation comparison dehazed images. process was learnable through provided pairs...
Motion blur is one of the most common degradation artifacts in dynamic scene photography. This paper reviews NTIRE 2020 Challenge on Image and Video Deblurring. In this challenge, we present evaluation results from 3 competition tracks as well proposed solutions. Track 1 aims to develop single-image deblurring methods focusing restoration quality. On 2, image are executed a mobile platform find balance running speed accuracy. targets developing video that exploit temporal relation between...
Image relighting is attracting increasing interest due to its various applications. From a research perspective, im-age can be exploited conduct both image normalization for domain adaptation, and also data augmentation. It has multiple direct uses photo montage aesthetic enhancement. In this paper, we review the NTIRE 2021 depth guided challenge.We rely on VIDIT dataset each of our two challenge tracks, including information. The first track one-to-one where goal transform illumination...
This paper reviews the first AIM challenge on bokeh effect synthesis with focus proposed solutions and results. The participating teams were solving a real-world image-to-image mapping problem, where goal was to map standard narrow-aperture photos same captured shallow depth-of-field by Canon 70D DSLR camera. In this task, participants had restore based only one single frame without any additional data from other cameras or sensors. target metric used in combined fidelity scores (PSNR SSIM)...
Dense video captioning is an extremely challenging task since accurate and faithful description of events in a requires holistic knowledge the contents as well contextual reasoning individual events. Most existing approaches handle this problem by first proposing event boundaries from then on subset proposals. Generation dense temporal annotations corresponding captions long videos can be dramatically source consuming. In paper, we focus generating temporally untrimmed aim to significantly...
Video deblurring remains a challenging task due to the complexity of spatially and temporally varying blur. Most existing works depend on implicit or explicit alignment for temporal information fusion, which either increases computational cost results in suboptimal performance misalignment. In this work, we investigate two key factors responsible quality: how fuse spatio-temporal from where collect it. We propose factorized gated attention module perform non-local operations across space...
This paper reviews the AIM 2019 challenge on extreme image super-resolution, problem of restoring rich details in a low resolution image. Compared to previous, this focuses an upscaling factor, ×16, and employs novel DIVerse 8K (DIV8K) dataset. report proposed solutions final results. The had 2 tracks. goal Track 1 was generate super-resolution result with high fidelity, using conventional PSNR as primary metric evaluate different methods. instead focused generating visually more pleasant...
Bokeh effect refers to the soft defocus blur of background, which can be achieved with different aperture and shutter settings in a camera. In this work, we present learning-based method for rendering such synthetic depth-of-field on input bokeh-free images acquired using ordinary monocular cameras. The proposed network is composed an efficient densely connected encoder-decoder backbone structure pyramid pooling module. Our leverages task-specific efficacy joint intensity estimation dynamic...
Image restoration is the task of recovering a clean image from degraded version. In most cases, degradation spatially varying, and it requires network to both localize restore affected regions. this paper, we present new approach suitable for handling image-specific spatially-varying nature in images by practically occurring artifacts such as rain-streaks, haze, raindrops motion blur. We decompose into two stages localization region-guided restoration, unlike existing methods which directly...
Image inpainting methods have shown significant improvements by using deep neural networks recently. However, many of these techniques often create distorted structures or blurry inconsistent textures. The problem is rooted in the encoder layers’ ineffectiveness building a complete and faithful embedding missing regions from scratch. Existing solutions like course-to-fine, progressive refinement, structural guidance, etc. suffer huge computational overheads owing to multiple generator...
Supervised networks address the task of low-light enhancement using paired images. However, collecting a wide variety low-light/clean images is tedious as scene needs to remain static during imaging. In this paper, we propose an unsupervised network context-guided illumination-adaptive norm (CIN). Inspired by coarse fine methods, in two stages. stage- I, pixel amplifier module (PAM) used generate estimate with overall improvement visibility and aesthetic quality. Stage- II further enhances...
This paper reviews the first-ever image demoireing challenge that was part of Advances in Image Manipulation (AIM) workshop, held conjunction with ICCV 2019. describes challenge, and focuses on proposed solutions their results. Demoireing is a difficult task removing moire patterns from an to reveal underlying clean image. A new dataset, called LCDMoire created for this consists 10,200 synthetically generated pairs (moire ground truth). The divided into 2 tracks. Track 1 targeted fidelity,...
Abstract Malaria is a major public health concern, causing significant morbidity and mortality globally. Monitoring the local population density diversity of vectors transmitting malaria critical to implementing targeted control strategies. However, current manual identification mosquitoes time-consuming intensive task, posing challenges in low-resource areas like sub-Saharan Africa; addition, existing automated methods lack scalability, mobile deployability, field-test validity. To address...
This paper reviews the NTIRE challenge on image colorization (estimating color information from corresponding gray image) with focus proposed solutions and results. It is first of its kind. The had 2 tracks. Track 1 takes a single as input. In 2, in addition to input image, some seeds (randomly samples latent are also provided for guiding process. operators were learnable through pairs training images. tracks 188 registered participants, 8 teams competed final testing phase.