Sunil Jaiswal

I am working as a computer vision researcher at K|Lens GmbH and also leading the computer vision-AI R&D team focussing on real-world computer vision problems in bringing ideas into reality from research prototypes to a product. My research interest includes:

  • Low-level vision : Real-world image & video super-resolution/deblurring/denoising for both single/multiview camera.
  • Geometric vision : Real-world multiview and monocular depth estimation, unsupervised video depth estimation, optical flow.
  • High-level vision : Real-world anamoly detection, defect detection/segmentation/classifications for both single/multiview camera.

    Prior to this, I obtained my Ph.D. in The Hong Kong University of Science & Technology (HKUST) in 2017 under the supervision of Prof. Oscar Au & Prof. Mattew McKay. During my Ph.D., I was a visiting research scholar in the VAI lab under Prof Klaus Mueller at Stony Brook University, USA and SUNY Korea in 2015/2016. I also visited Technicolor R&D, France as a research internee and awarded "Best Research and Innovation Award" of 2017. In 2012, I received an undergraduate degree from LNMIIT, India in the ECE department, where I worked with Prof. Anil Tiwari & visited CBIA, Czech Republic.

    Email  /  CV  /  Google Scholar  /  LinkedIN

  • profile photo


    Experience

    I am fortunate enough to work with some great minds:

  • Jan'2018 - Present : Computer Vision Researcher at K|Lens, GmbH, Germany
  • Sep'2012 - Dec'2017 : Ph.D. Candidate at HKUST, Hong Kong under the supervision of Prof. Oscar Au & Prof. Mattew McKay
  • Apr'2015 - Mar'2016 : Visiting Scholar at Stony Brook University, USA & SUNY Korea under the supervision of Klaus Mueller
  • Dec'2015 - Dec'2017 : Worked with Prof. Lu Fang, Tsinghua University, China
  • June'2017 - Oct'2017 : Research internee at Technicolor R&D, France under the supervision of Dr. F. Galpin, Dr. R. Fabien
  • Jun'2011 - Aug'2011 : Research internee at CBIA, Czech Republic under the supervision of M. Kozubek, D. Svoboda
  • May - Aug'2010 & 2012 : Reseach internee at IIT Jodhpur, INDIA under the supervision of Prof. Anil Tiwari


  • Selected Publication

    Expanding Synthetic Real-World Degradations for Blind Video Super Resolution
    Mehran Jeelani*, Sadbhawna Thakur*, Noshaba Cheema, Klaus Illgner, Philipp Slusallek, Sunil Jaiswal (*equal contribution)
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2023
    Paper

    This work shows how varied random degradations can contribute to learning an effective VSR model, especially for real-world video artifacts.

    High-Resolution Synthetic RGB-D Datasets for Monocular Depth Estimation
    Aakash Rajpal, Noshaba Cheema, Klaus Illgner, Philipp Slusallek, Sunil Jaiswal
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2023
    Paper, HRSD_DATASET

    In this work, we generate a high-resolution synthetic depth dataset (HRSD) which contains 100,000 color images and corresponding dense ground truth depth maps.

    Leveraging Multi-view Data for Improved Detection Performance: An Industrial Use Case
    Faranak Shamsafar, Sunil Jaiswal, Benjamin Kelkel, Kireeti Bodduna, Klaus Illgner-Fehns
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2023
    Paper

    In this work, We introduce a novel multi-view dataset with semi-automatic ground-truth data, which results in significant labeling resource savings.

    Edge-aware Consistent Stereo Video Depth Estimation
    Elena Kosheleva,*, Sunil Jaiswal*, Faranak Shamsafar, Noshaba Cheema, Klaus Illgner, Philipp Slusallek (*equal contribution)
    Arxiv'2023
    Paper

    In contrast to the naive method of estimating depths from images, a more sophisticated approach uses temporal information, thereby eliminating flickering and geometrical inconsistencies. We propose a consistent method for dense video depth estimation; however, unlike the existing monocular methods, ours relates to stereo videos.

    Context Region Identification based Quality Assessment of 3D Synthesized Views
    Sadbhawna Thakur*, Vinit jakhetiya, Badri N. Subudhi, Sunil Jaiswal, Leida Li, Weisi Lin
    IEEE Transactions on Multimedia, 2022
    Paper

    In this work, we propose a new and efficient quality assessment algorithm based upon the variation in the depth of 3D synthesized and reference views.

    A Comparative Study of PnP and Learning Approaches to Super-Resolution in a Real-World Setting
    Samim Zahoor Taray, Sunil Jaiswal, Shivam Sharma, Noshaba Cheema, Klaus Illgner-Fehns, Philipp Slusallek, Ivo Ihrke
    GCPR'2021
    Paper

    The applicability to real-world images is limited and expectations are often disappointed when comparing to the performance achieved on synthetic data. For improving on this aspect, we investigate and compare two extensions of orthogonal popular techniques, namely plug-and-play optimization with learned priors, and a single end-to-end deep neural network trained on a larger variation of realistic synthesized training data, and compare their performance with special emphasis on model violations. We observe that the end-to-end network achieves a higher robustness and flexibility than the optimiza- tion based technique.

    Method for digital image processing
    Klaus Illgner-Fehns, Samim Zahoor Taray, Sunil Jaiswal,
    US Patent App. 18/036,807
    Paper
    Lightfield imaging for industrial applications.
    Klaus Illgner-Fehns, John Restrepo, Sunil Jaiswal, Ivo Ihrke
    SPIE Future Sensing Technologies, 2020
    Paper

    Lightfield imaging systems were brought to market for consumer and professional media recording. But so far, this technology is less known for applications in the industrial space. The unique optical concept developed by KLens allows to capture multiple perspectives of a scene with a single exposure as regular colour images on the camera sensor.

    Kernel-ridge regression-based quality measure and enhancement of three-dimensional-synthesized images.
    Vinit Jakhetiya, Ke Gu, Sunil Jaiswal, Trisha Singhal, Zhifang Xia
    IEEE Transactions on Industrial Electronics, 2020
    Paper

    In this article, we propose an efficient joint image quality assessment and enhancement algorithm for the 3-D-synthesized images using a global predictor, namely, kernel ridge regression (KRR).

    A Prediction Backed Model for Quality Assessment of Screen Content and 3D synthesized Images.
    Vinit Jakhetiya, Gu Ke, Weisi Lin, Qiaohong Li, Sunil Jaiswal
    IEEE Transactions on Industrial Informatics, 2020
    Paper

    In this paper, we address problems associated with free-energy-principle-based image quality assessment (IQA) algorithms for objectively assessing the quality of Screen Content (SC) and three-dimensional (3-D) synthesized images and also propose a very fast and efficient IQA algorithm to address these issues.

    Method and apparatus for encoding a picture block
    Fabrice LELEANNEC, Franck Galpin, Fabien Racape, Sunil Jaiswal
    US Patent 16/772,037
    Paper
    Texture-based partitioning decisions for video compression
    Fabrice LELEANNEC, Franck Galpin, Fabien Racape, Sunil Jaiswal
    US Patent 16/771,094
    Paper
    Deep learning based image partitioning for video compression
    Fabrice LELEANNEC, Franck Galpin, Sunil Jaiswal, Fabien Racape
    US Patent App. 16/771,105
    Paper
    CNN-Based Driving of Block Partitioning for Intra Slices Encoding.
    Franck Galpin, Fabien Racapé, Sunil Jaiswal, Philippe Bordes, Fabrice Le Léannec Edouard François
    Data Compression Conference (DCC), 2019
    Paper

    An encoding approach based on Convolutional Neural Networks is explored to partly substitute classical heuristics-based encoder speed-ups by a systematic and automatic process. The solution allows controlling the trade-off between complexity and coding gains, in intra slices, with one single parameter.

    Just noticeable difference for natural images using RMS contrast and feed-back mechanism.
    Vinit Jakhetiya, Weisi Lin, Sunil Jaiswal, Gu Ke, Sharath C. Guntuku
    Elsevier Neurocomputing, 2018
    Paper

    We propose the first pixel-based JND algorithm that includes a very important component of the human vision, namely CS by measuring RMS contrast. This RMS contrast is combined with LA and CM to form a comprehensive pixel-domain model to efficiently estimate JND in the low frequency regions. For high frequency regions (edge and texture), a feedback mechanism is proposed to efficiently alleviate the over- and under-estimation of CM.

    Adaptive Multispectral Demosaicking Based on Frequency Domain Analysis of Spectral Correlation.
    Sunil Jaiswal, Lu Fang, Vinit Jakhetiya, Jiahao Pang, Klaus Mueller, Oscar Au
    IEEE Transactions on Image Processing, 2017
    Paper

    Color filter array (CFA) interpolation, or three-band demosaicking, is a process of interpolating the missing color samples in each band to reconstruct a full color image. In this paper, we are concerned with the challenging problem of multispectral demosaicking, where each band is significantly undersampled due to the increment in the number of bands.

    Maximum a Posterior and Perceptually Motivated Reconstruction Algorithm: A Generic Framework.
    Vinit Jakhetiya, Weisi Lin, Sunil Jaiswal, Sharath C Guntuku Oscar Au
    IEEE Transactions on Multimedia, 2017
    Paper

    In this paper, we propose an efficient perceptually motivated and maximum a posterior (MAP)-based generic framework for image reconstruction. This can be applied to several image/video processing applications, where there is a necessity to improve reconstruction accuracy and suppress visible artifacts, such as denoising, deinterlacing, interpolation, de-blocking of Jpeg/Jpeg-2000, and demosaicing.


    Last updated on January 11, 2024 | Thanks Dr. Jonathan T. Barron for this awesome template.