STUDENTS
Vanshaj Chowdhary
Vanshaj is a graduate student pursuing a Masters in Computer Vision degree at Carnegie Mellon University. Previously, he worked as a Software Engineer at Amazon Robotics, building autonomous vehicles for warehouse applications. He has extensive experience developing and deploying core perception components to production systems.
Sahiti Yerramilli
Sahiti is a first-year MSCV student at Carnegie Mellon University. She completed her undergraduate degree in Computer Science from the National Institute of Technology, Warangal (NITW), India. Previously, she was a Software Engineer at Microsoft where she worked in the field of Mixed Reality for Microsoft Hololens. During her undergraduate studies, she also worked under Dr. Chakravarthy Bhagvati on robust multi-frame super-resolution.
ADVISORS
Jeff Schneider
Dr. Schneider’s research interests are in all areas of machine learning and data mining. He has over 15 years of experience developing, publishing, and applying machine learning
algorithms in government, science, and industry. He has over a hundred publications and has given numerous invited talks and tutorials on the subject.
Christoph Mertz
Dr. Mertz received his PhD. in Physics from Arizona State University. After continuing his nuclear physics research for another year, he joined the CMU Robotics Institute research staff. There he developed surround sensing and object detection systems for intelligent vehicles and search and rescue robots. He worked on projects sponsored by ARL, DOT, FTA, PennDOT, NIJ, DARPA, NSF, and several companies. In 2016, Dr. Christoph Mertz became Co-Founder and Chief Scientist at RoadBotics while remaining the Principal Project Scientist at CMU, where he has conducted academic research for over 20 years.
Project Responsibilities
Vanshaj and Sahiti worked together to formulate the problem statement and methodology under the advice of Dr. Christoph Mertz and Dr. Jeff Schneider. During the Spring semester, Vanshaj focused on experimenting with existing stereo-based methods and evaluating them on different datasets. Sahiti focused on exploring and applying domain adaptation for stereo-based purposes. During the Fall semester, Vanshaj and Sahiti explored extending Monodepth2 to our use case. Vanshaj worked closely with the NREC dataset, and Sahiti worked on the KAIST dataset.