Posts by Collection

portfolio

publications

Learning based Blur Detection and Segmentation

Published in ICIP 2018, 2018

We present a robust two-level architecture for blur-based segmentation of a single image. First network is a fully convolutional encoder-decoder for estimating a semantically meaningful blur map from the full-resolution blurred image. Second network is a CNN-based classifier for obtaining local (patch-level) blur-probabilities. Fusion of the two network outputs enables accurate blur-segmentation using Graph-cut optimization over the obtained probabilities. We also show its applications in blur magnification and matting.

Bringing Alive Blurred Moments

Published in CVPR 2019, 2019

We present a solution for the goal of extracting a video from a single motion blurred image to sequentially reconstruct the clear views of a scene as beheld by the camera during the time of exposure. We first learn motion representation from sharp videos in an unsupervised manner through training of a convolutional recurrent video autoencoder network that performs a surrogate task of video reconstruction. Once trained, it is employed for guided training of a motion encoder for blurred images. This network extracts embedded motion information from the blurred image to generate a sharp video in conjunction with the trained recurrent video decoder. As an intermediate step, we also design an efficient architecture that enables real-time single image deblurring and outperforms competing methods across all factors: accuracy, speed, and compactness. Experiments on real scenes and standard datasets demonstrate the superiority of our framework over the state-of-the-art and its ability to generate a plausible sequence of temporally consistent sharp frames.

talks

3D Reconstruction System

Built a system to reconstruct a 3D scene given images captured from an internally calibrated camera. Essential matrices for each pair of images was found out. Performed stepwise gluing to obtain camera positions and point cloud. Project was part of a course on 3D Computer Vision

MNIST Digit Classification

Training a FCN from scratch using Numpy for MNIST digit classification. Support for various optimizers, loss functions, layers. Achieved a 98.5% accuracy. (Github)

Indoor Navigation

Conceptualized and built a system to navigate a person to a room inside a building. Used Google’s speech API for voice recognition and generation. Implemented on the RaspberryPi platform. (Github)

Robot Navigation Using Kinect

Developed a system to control a robot autonomously using Kinect. Location of the robot is found out using Trilateration thorugh the help of predefined markers. Controlled using an Arduino Microcontroller.

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post.