About

I conduct research in computer vision with specific focus on object detection and recognition. Till now, I have worked on various different projects for benchmarking applications of computer vision. I prefer to follow an intuitive approach to identify and solve pre-existing problems using mathematical and statistical foundations. I thoroughly believe that there will come a day when all we will require to detect, identify and describe objects will be a mobile phone equipped with a camera. This is the world that I look forward to and I want to add to it as much as I can as a researcher.
Read more about my research here.
On the more personal side of my life, I am an avid reader and gamer. I have been an editor for my college magazine and have a knack for writing as well. I hope to publish a book soon.
I was involed with the Computer Vision Group at my undergrauate institute L.D. College of Engineering, Ahmedabad and was one of it's founding members.

Publications

SAF-BAGE: Salient Approach for Facial Soft-Biometric Classification - Age, Gender, and Facial Expression
Ayesha Gurnani, Kenil Shah, Vandit Gajjar, Viraj Mavani, Yash Khandhediya
(Authors Contributed Equally)
WACV 2019 (to appear)
ViS-HuD: Using Visual Saliency to Improve Human Detection with Convolutional Neural Networks
Vandit Gajjar, Yash Khandhediya, Ayesha Gurnani, Viraj Mavani, Mehul Raval
CVPR 2018
Facial Expression Recognition using Visual Saliency and Deep Learning
Viraj Mavani, Krishna P Miyapuram, Shanmuganathan Raman
ICCV 2017

Pet Projects

Anno-Mage

A Semi Automatic Image Annotation Tool to help create object detection datasets faster.

Anno-Mate

A Tool which helps you to convert popular image annotation formats to other popular image annotation formats.

Emotion-Recognition

My emotion recognition code incorporates a detection recognition pipeline to understand emotions in real-time.

Object-Detection-Demo-GUI

A GUI demo using convolutional neural network and single shot detector to identify 20 objects in images.

Wind-Speed-Prediction

MATLAB implementation to predict wind speed on the basis of temperature and season. The London Meteorological Data was used for training.