AJJEN JOSHI
PhD student @ Boston University
Creative @ ajjenjoshi.com

ABOUT


I am currently a PhD student working under the supervision of Dr. Margrit Betke and Dr. Stan Sclaroff with the Image and Video Computing research group at Boston University's Department of Computer Science. I attained a Master of Science degree in Computer Science from Boston University in 2014, and a Bachelor of Arts degree from Connecticut College in 2012 where I majored in Computer Science and Architectural Studies and completed a certificate program in Arts and Technology. I have had the opportunity to intern at Disney Research (Summer 2015) and Adobe Creative Technologies Lab (Summer 2016).

My broad research interests lie in computer vision, machine learning, and human computer interaction. Particularly, I am interested in the analysis of spatio-temporal human signals. In my spare time, I do freelance work in art and photography. Some of my work can be seen at ajjenjoshi.com.

RESEARCH



DEEPTWEEN: A DATA-DRIVEN APPROACH TO AUTOMATIC INBETWEENING IN HANDDRAWN ANIMATION (current)



Ajjen Joshi, Soumya Ghosh, Margrit Betke, Stan Sclaroff Hanspeter Pfister
PERSONALIZING GESTURE RECOGNITION USING HIERARCHICAL BAYESIAN NEURAL NETWORKS
Conference on Computer Vision and Pattern Recognition (CVPR), 2017. Poster.

[Paper] | [Supplement]



Ajjen Joshi, Soumya Ghosh, Margrit Betke, Hanspeter Pfister
HIERARCHICAL BAYESIAN NEURAL NETWORKS FOR PERSONALIZED CLASSIFICATION
Neural Information Processing Systems (NIPS) Workshop on Bayesian Deep Learning, 2016. Poster.

[Paper] | [Poster] | [Spotlight]



Ajjen Joshi, Linda Tickle-Degnen, Sarah Gunnery, Terry Ellis, Margrit Betke
PREDICTING ACTIVE FACIAL EXPRESSIVITY IN PEOPLE WITH PARKINSON'S DISEASE
International Conference on Pervasive Technologies Related to Assistive Environments (PETRA), 2016. Oral.

[Paper]


Ajjen Joshi, Camille Monnier, Margrit Betke, Stan Sclaroff
COMPARING RANDOM FOREST APPROACHES TO SEGMENTING AND CLASSIFYING GESTURES
Elsevier Image and Vision Computing (IMAVIS), 2016.

[Paper]


Andrew Kurauchi, Wenxin Feng, Ajjen Joshi, Carlos Morimoto, Margrit Betke
EYESWIPE: DWELL-FREE TEXT ENTRY USING GAZE PATHS
ACM Conference on Human Factors in Computing Systems (CHI), 2016. Oral.

[Paper] | [Video]


Huy Le, Ajjen Joshi, Margrit Betke
B3.JS: A LIBRARY FOR INTERACTIVE VIRTUAL REALITY WEB 3D GRAPHS
IEEE Virtual Reality (VR), 2016. Research Demo.

[Paper] | [Video]


Ajjen Joshi, Camille Monnier, Margrit Betke, Stan Sclaroff
A RANDOM FOREST APPROACH TO SEGMENTING AND CLASSIFYING GESTURES
IEEE International Conference on Automatic Face and Gesture Recognition (AFGR), 2015. Oral.

[MS Thesis] | [Paper] | [Video]


Ajjen Joshi, Bridget Baird, Ozgur Izmirli
DEVELOPING A TOOL FOR DANCE MOTION SYNTHESIS
Connecticut College Biennial Symposium on Arts and Technology, 2012. Oral.

[Paper]

PROJECTS

GESTURE RECOGNITION FROM GLOVE SENSORS (2015)


Built a prototype of a gesture recognition system trained to recognize gestures performed by using a glove equipped with various sensors during my internship at Disney research.

#Python, scikit-learn, IMU data


VICTORY OVER THE SUN (2015)


Created projections for a rendition of the futurist opera: Victory Over the Sun. Wrote scripts creating generative animations and used the Kinect to record and animate dancers.

#Processing, Kinect


KEY EYE (2015)


A prototype of a hands-free eye typing system (created by a 2-person team). Designed for people with motor impairments, users typed words using their eyegaze with a wearable eye-tracker.

#Python, scikit-learn, pupil-labs eyetracker, Video


IMPLEMENTATION OF A RAY TRACER IN JAVA (2013)


A ray tracer with support for quadric surfaces (spheres, ellipsoids, cylinders, boxes), 5 directional light sources, lighting models with ambient, diffuse and specular terms, as well as effects caused by shadows, reflections and refractions.

#Java, OpenGL


PAPER PIANO (2012)


An openCV app (created by a 2-person team) that allowed users to play an 8-key paper piano. Computer vision algorithms were user to detect keys, segment user hands and determine which keys were being pressed.

#C++, OpenCV


WINCHESTER KITCHEN (2011)


A web app written using the LAMP (Linux, Apache, MySQL, PHP) stack that allowed users to select food available in a student run kitchen and place an order for pickup or delivery.

#HTML, CSS, Javascript, LAMP stack


TEACHING


ARTIFICIAL INTELLIGENCE (Spring 2016)
Boston University CS440 (Undergraduate course on artificial intelligence)

IMAGE AND VIDEO COMPUTING (Fall 2014)
Boston University CS585 (Graduate level course on computer vision)

INTRODUCTION TO APPLICATION PROGRAMMING (Fall 2013)
Boston University CS108 (Introductory course on computer programming)

3D GAME DESIGN (Spring 2013)
Boston University STEM Academy (Introductory course on game design for high school students)

MULTIMEDIA PROCESSING (Fall 2011)
Connecticut College CS218 (Introductory course on image and audio processing)

ANIMATION AND MOTION CAPTURE WORKSHOP (Winter 2010, 2011)
Connecticut College Center for Arts and Technology (Workshop on 3d animation and motion capture)

TESTIMONIALS

"Ajjen was the best teaching fellow I have had at BU thus far. He is very knowledgeable in all things computer vision and was helpful when explaining topics." -CS585 student, Fall 2014

"Ajjen is extremely qualified for this position and he clearly demonstrated it by helping students outside of class with projects he is not familiar with. He is extremely helpful and definitely deserves a good recommendation." -CS108 student, Fall 2013


CONTACT