Shyam Nandan Rai

Shyam Nandan Rai

MS by Research in CV/ML

CVIT, IIIT-Hyderabad


Hi!! I am Shyam Nandan Rai, working as a research assistant at CVIT, IIITH. Currently, I am involved in the vision and safety group at CVIT, working to improve and make autonomous systems more reliable. In detail, I address the problem of minimizing the rain streaks in the images by a few-shot self-supervised method and use restored images to improve computer vision tasks such as pedestrian detection that are critical to autonomous systems. My current research interest lies explicitly in Semantic Segmentation, Image Restoration in adverse weather, GANs, and problems related to the imbalance in large-scale datasets.

I completed my Masters (MS) at IIIT Hyderabad, where I was under the joint supervision of Prof. C.V. Jawahar, Prof. Vineeth N Balasubramanian, and Dr. Anbumani Subramanian on the problems of image restoration in adverse weather conditions. Previously, I was an undergrad honors student at IIIT, Sricity, where I worked on the problems related to Cartoon Image Understanding in collaboration with Prof C.V. Jawahar and Prof. Anand Mishra.


  • Computer Vision
  • Machine Learning
  • Image Restoration
  • Generative Adverserial Network


  • M.S. (Research) in Computer Science, 2020

    CVIT, International Institute of Information Technology, Hyderabad

  • B.Tech (Hons.) in Electronics and Communication Engineering, 2017

    Indian Institute of Information Technology, Sri City


IIIT-CFW: A Benchmark Database of Cartoon Faces in the Wild (ECCVW 2016)

The IIIT-CFW is database for the cartoon faces in the wild. It is harvested from Google image search.
IIIT-CFW: A Benchmark Database of Cartoon Faces in the Wild (ECCVW 2016)

Learning To Generate Atmospheric Turbulent Images (NCVPRIPG 2020)

We train a generative adversarial network which outputs an atmospheric turbulent image by utilizing less computational resources than traditional methods.
Learning To Generate Atmospheric Turbulent Images (NCVPRIPG 2020)

Munich-to-Dubai: How far is it for Semantic Segmentation ? (WACV 2020)

Cities having hot weather conditions results in geometrical distortion, thereby adversely affecting the performance of semantic segmentation model. In this work, we study the problem of semantic segmentation model in adapting to such hot climate cities.
Munich-to-Dubai: How far is it for Semantic Segmentation ? (WACV 2020)

Spatial Feedback Learning to Improve Semantic Segmentation in Hot Weather (BMVC 2020)

High-temperature weather conditions induce geometrical distortions in images which can adversely affect the performance of a computer …
Spatial Feedback Learning to Improve Semantic Segmentation in Hot Weather (BMVC 2020)



Detecting misalignment in CAD images

Developed an interactive computer vision application using Qt and Opencv, to detect misalignment in CAD Images.

Facial Expression Recognition

We used pre-trained model VGG Face for extracting features and used SVM for classification with different kernels. In addition, we used a unified model which fuses the CNN features and HOG feature giving higher accuracy than other models.

Gender Identification from Facial Images

Implemented different feature based methods to identify gender from facial images. Extended the method to cross modal gender identification between the real face and its cartoon and caricature modalities.

Image Quality Assessment

Method for designing high level features for photo quality assessment.

LRR Network for Semantic Segmentation

Multi-scale architecture based on a laplacian pyramid approach to improve semantic segmentation.

Multi Agent Diverse Generative Adversarial Networks

Implementation of MAD-GAN which addresses the problem of mode collapse.

Reading Comprehension

Posed reading comprehension as a sentence classification task. Instead of sequential models, we used CNN models for classification and extended to a siamese variation using contrastive loss.

Relationship between Music & Personality

Objective of this project is to find a strong correlation between personality of an individual and their music preferences.

Twin Auxiliary Classifier GAN

Implementation of Twin Auxiliary Classifiers Generative Adversarial Net (TAC-GAN) which significantly improves the diversity of class-conditional image generation.



AI/ML mentor

MLL Lab-IIITH|Talentsprint

Oct 2017 – Present Hyderabad, India

Research Assistant

CVIT, IIIT-Hyderabad

Jul 2016 – Present Hyderabad, India

Computer Vision Intern

RTC, Robert Bosch

May 2016 – Jul 2016 Bangalore, India