Krishnendu Chaudhury

Krishnendu Chaudhury is an American Technology leader (deep learning, computer vision), inventor, and author. He is the co-founder of Drishti Technologies in Palo Alto, CA. Currently, he is serving as the Chief Technology officer (CTO) of the company. Chaudhury is also the author of the book Mathematics and Architectures of Deep Learning. He has 34 patents to his name and several publications in the field of computer vision and deep learning. Throughout his career, he has worked at Adobe Systems, Google, and Flipkart.

Early life and education
In 1994, he completed his Ph.D. in computer science in computer vision and image processing from the University of Kentucky.

Career
In 1995, Chaudhury started working for Adobe Systems as a senior computer scientist. He worked with Adobe for ten years working on Advanced Technology Group and Postscript Core Technology, working in different locations, including San Francisco and San Jose.

In 2005, he joined Google. He worked on image processing and computer vision projects. He worked on several projects that revolved around machine learning and computer vision, including Google's newspaper search feature, which was launched in 2008 on Google's 10th birthday. The product allows readers to go through newspaper archives from hundreds of newspapers and thousands of issues from the 1800s and early 1900s. Chaudhury also managed Google's version of image compression called "WebP," which was launched in October 2010. He worked on the auto-rectification of photos at Google Photos, devising a complex mathematical way to restore the parallelism of lines inside an image. He also worked on an early version of a face recognition-based login for Android. After 10 years of service, he left Google and joined Flipkart.

In 2015, Chaudhury joined Flipkart as a Principal Scientist and Head of Image Sciences, handling projects related to computer vision and deep learning. He worked on deep learning in the e-commerce visual search and recommendation systems. Chaudhury led the computer vision and image processing technology team across Flipkart's platforms. His team led the first visual recommendation engine, which would recommend visually similar products. His notable works for Flipkart include Deep Learning based Visually Similar recommendations for Flipkart: A new Convolutional Neural Net (CNN) architecture for visual feature embeddings, and Region CNN-based e-commerce product localization on wild images: Identifying Flipkart apparel similar to the ones worn by models/celebrities/friends in wild images.

In 2017, he co-founded Drishti Technologies in Palo Alto, CA, with Prasad Akella and serial entrepreneur Ashish Gupta. Drishti Technologies works on AI-powered video analytics and video traceability. Drishti is a Silicon Valley AI startup funded by VCs including Andreessen Horowitz, Emergence Capital, and Benhamou Global Ventures. Chaudhury is also on the Technical Advisory Committee of Benhamou Global Ventures since 2017. He holds the position of CTO at Drishti Technologies, working on Applying Deep Learning and Computer Vision.

Selected publications

 * Loss and Gain of Optical Couplers with Collision Detection for CSMA/CD based LANs, A. Das, Krishnendu Chaudhury, IEEE International Conference on Communication Systems, ICCS-89, Singapore, Oct. 12-16, 1989
 * A Parallel Algorithm for 3D Point Pattern Matching, Krishnendu Chaudhury, R. Mehrotra, IEEE International Conference on Systems, Man and Cybernetics, Charlottesville, Virginia, Oct. 13-16 1991.
 * Optical Flow from an Extended Frame Sequence, Krishnendu Chaudhury, R. Mehrotra, International Conference on Robotics and Automation, Atlanta, Georgia, May 2–7, 1993.
 * "Detecting 3D Flow", Krishnendu Chaudhury, R. Mehrotra, International Conference on Robotics and Automation, San Diego, California, May 8–13, 1994.
 * "Optical Flow Estimation using Smoothness of Intensity Trajectories," Krishnendu Chaudhury, R. Mehrotra, CVGIP: Image Understanding journal, pp. 230–244, vol. 60, No. 2, September 1994.
 * "A Trajectory-based Computational Model for Optical Flow Computation," Krishnendu Chaudhury, R. Mehrotra, IEEE Transactions on Robotics and Automation journal, pp. 733–741, vol. 11, No. 5, Oct 1995
 * "Pose Estimation in Automated Visual Inspection using Genetic Algorithms," S. Hati, K. Chadhury, et al., International Journal of Neural Systems, Aug 1, 2006.
 * "Auto-Rectification of User Photos," Krishnendu Chaudhury, Stephen DiVerdi, Sergey Ioffe, proc. of IEEE International Conference on Image Processing, ICIP 2014, Oct 27-30, 2014, Paris.
 * "Auto-Rectification of User Photos," Krishnendu Chaudhury, Stephen DiVerdi, Sergey Ioffe, proc. of IEEE International Conference on Image Processing, ICIP 2014, Oct 27-30, 2014, Paris.

Selected patents

 * Chaudhury has more than 34 patents to his name in the field of computer vision and deep learning.
 * "Almost Unsupervised Cycle and Action Detection," K. Chaudhury, A.Ashok, S. Narumanchi, D. Shankar, A. Mehra, Drishti Technologies, US Patent 20210216777
 * "Deep Learning Cycle Detection from the video," K. Chaudhury, A.Ashok, S. Narumanchi, D. Shankar, Drishti Technologies, R. Jain, DRIL-P0013, US Patent 20210117684
 * "Traceability Systems and Methods," P.Akella, K. Chaudhury, A.Ashok, S. Narumanchi, D. Shankar, Drishti Technologies, US Patent 10890898B2
 * "Automatic Rectification of Distortions in images," Krishnendu Chaudhury, Stephen Diverdi, US Patent 9064309.
 * "Edge Aware Smoothing in Images," US Patent US8983188, Grayson Lang, Krishnendu Chaudhury
 * "Session Based Character Recognition for Document Reconstruction," D. Petrou, Krishnendu Chaudhury, S. Goschin, M. Bridges, US Patent 9424668.
 * "Liveness Detection in Face Recognition-based Authentication Systems," Krishnendu Chaudhury, Avani Devarasetty, U.S Patent US8856541.
 * "Automatic extraction of character ground truth data from images," Alessandro Bissacco, Krishnendu Chaudhury, U. S. Patent 8755595.
 * "Systems and Methods for Line Balancing," P.Akella, K. Chaudhury, et al., US Patent 11054811
 * "Handheld or Wearable Document Scanner for Google," Krishnendu Chaudhury, Lu Chen, David Petrou, Blaise Aguerra, US Patent 9852348.
 * "Segmenting Printed Media Pages into Articles," Krishnendu Chaudhury, A. Jain & S. Thirthala, U.S Patent 8290268, Google
 * "Identifying a Front Page in Media Material," Krishnendu Chaudhury et al., U.S Patent 8218913, Google
 * "Method and Apparatus for Enhancing Object Boundary Precision in Images," Krishnendu Chaudhury, Ashutosh Kulshreshtha, U.S patent 07813582, Google.
 * "Digital Image Archiving in Mobile Device System," Krishnendu Chaudhury, Ashutosh Garg, Arvind Saraf, Prasenjit Phukan, U.S Patent 7986843, Google.
 * "Complexity Based Adaptive Tiling for Transparency Flattener," Krish Chaudhury, Dejan Markovic, U.S Patent 7385727, Adobe Systems.
 * "A Cubic B Spline Intensity Re-sampling Based Method for Automatic Image Feature Embedding," Krish Chaudhury, Dejan Markovic, U.S. Patent 7734118, Adobe Systems.
 * "Scaling of Raster Images without Edge Blurring," Krish Chaudhury, Dejan Markovic, U.S. Patent 7817871, Adobe Systems.
 * "Author Signature for Legal Purposes," Jim Pravetz, Krish Chaudhury, Sunil Agrawal, U.S. Patent 7774608, Adobe Systems).
 * "Document Modification Detection and Prevention," Krish Chaudhury, Jim Pravetz, Sunil Agrawal, U.S. Patent 7735144, Adobe Systems.
 * "Document Digest Allowing Selective Changes to a PDF Document," Krish Chaudhury, Jim Pravetz, International Patent, Adobe Systems, PCT/US2004/015116).
 * "Dynamic Enabling of Functionality in Electronic Document Readers," Krish Chaudhury, Jim Pravetz, U.S. Patent 7278168, Adobe Systems.

Book
In 2020, Chaudhury authored Math and Architectures for Deep Learning which was published by Manning Publications, New York.

Awards and recognition
In 2022, Chaudhury was included in Best Startup magazine's list of "California's 101 Top CTO's in the Machine Learning Space ".