Tarasha Khurana
I am an MS in Robotics student at The Robotics Institute, Carnegie Mellon University, advised by Prof. Deva Ramanan. I am currently working on reasoning about occlusions in 3D for application in amodal detection and consequently, tracking. Previously, I worked as a Research Associate at Staqu Technologies on AI problems in vision and speech. During my undergrad, I gained a premium research experience at the Computer Vision and Machine Learning Lab at IIIT-Delhi under the mentorship of Dr. Chetan Arora, who is now at IIT Delhi. I obtained a B.E. in Computer Engineering from University of Delhi where I developed a depth in vision under the guidance of Dr. Anand Gupta.

[05/20] The TAO Dataset manuscript has now been released on arXiv!
[04/01] I'll be at Argo AI for the summer as a Research Intern!
[02/10] Our workshop titled, 'Object Tracking and its Many Guises' was accepted at ECCV 2020!
[01/29] I am reviewing for ICML 2020!
Academic Research
TAO: A Large-Scale Benchmark for Tracking Any Object
Achal Dave, Tarasha Khurana, Pavel Tokmakov, Cordelia Schmid, Deva Ramanan
[abstract] [arXiv] [project page] [workshop]
Exploiting Texture Cues for Clothing Parsing in Fashion Images
Tarasha Khurana, Kushagra Mahajan, Chetan Arora, Atul Rai
25th IEEE International Conference on Image Processing (ICIP) '18.
[abstract] [pdf] [poster] [Xplore]
Pose Aware Fine Grained Visual Classification Using Pose Experts
Kushagra Mahajan, Tarasha Khurana, Ayush Chopra, Isha Gupta, Chetan Arora, Atul Rai
25th IEEE International Conference on Image Processing (ICIP) '18.
[abstract] [pdf] [poster] [Xplore]
Prediction of Lung Cancer from Low Resolution Nodules in CT-Scan Images by using Deep Features
Anand Gupta, Sagorika Das, Tarasha Khurana, Kamakshi Suri
IEEE's 7th International Conference on Advances in Computing, Communications & Informatics (ICACCI) '18.
[abstract] [Xplore]
Table Detection and Metadata Extraction in Document Images
Anand Gupta, Devendra Tiwari, Tarasha Khurana, Sagorika Das
Springer's 2nd International Conference on Smart Innovations in Communications & Computational Sciences (ICSICCS) '18.
[abstract] [Springer]
Some Special Characteristics of Tracks and Intensity of Typhoons over the Western North Pacific
Tarasha Khurana, S.K. Bhattacharya, S.D. Kotal
United Nations ESCAP/World Meteorological Organization Typhoon Committee's Tropical Cyclone Research and Review Journal (TCRR), Volume 6, No. 3-4.
[abstract] [TCRR]
Reading Indian Road Signboards
Tarasha Khurana, Isha Gupta, Chetan Arora
Industrial Research
Earth Observation
Earth Observation involves localisation and sub-classification of target objects of size smaller than 10px by 15px from huge satellite images of size 50,000px by 50,000px.
Person Video Analysis
Person video analysis involves processing temporal information to perform multi-person tracking, gait recognition and audio-video lipreading on occluded and low-quality camera streams.
Adversarial Example Generation
Assessment of deep models against adversarial attacks involves generating adversarial examples using white-box and black-box methods.