Learn OpenCV: Insights from a Computer Vision and Deep Learning Entrepreneur and Enthusiast
The OpenCV (Open-Source Computer Vision) Library is, as many of you likely already know, a key enabler for the practical deployment of computer vision technology, the fundamental mission of the Embedded Vision Alliance. The Alliance website contains a rich collection of OpenCV downloads, documentation, tutorial videos and other resources, as does (of course) the OpenCV Foundation's own website. A number of other useful independent sources of OpenCV information also exist, and one of the newest is Learn OpenCV, the brainchild of computer vision and deep learning developer Satya Mallick.
The Alliance recently contacted Mallick to further understand his background, his motivations for launching the site, and his vision for its continued evolution. The following lightly edited version of the interview transcript provides numerous insights; post your additional questions and thoughts in the comments section for Mallick to see and respond to.
What initially attracted you to computer vision, and what has kept you interested in it over the years?
As a starry-eyed undergraduate student who loved Isaac Asimov, I was enamored by the idea of machines that could see and think. I was not sure if machines would ever visually interpret their surroundings like humans do, but I had no doubt that the attempt was worth the effort.
What has kept me interested in the field, over more than a dozen years to date, is the way that the computer vision community has tackled difficult problems and brought robust solutions to the real world that were found only in the realm of science fiction even just a few years ago. This field is full of dreamers, which is a good thing!
I am also fascinated by the fact that computer vision solutions often bring people of different disciplines -- computer scientists, engineers, mathematicians, programmers, visual artists, and biologists, to name a few -- together as a collaborative team. Finally, this is a field of research that promises to solve real-world problems, and not in the some distant future, but today.
What is your educational background?
I have a Ph.D. in Computer Vision and Machine Learning from the University of California, San Diego. Prior to that, I obtained Bachelors Degree of Technology in Electrical Engineering from the Indian Institute of Technology in Kharagpur.
What is your professional background, what companies are you currently involved in, and what technologies and products are they involved in?
TAAZ is a company I started with my Ph.D. advisor Dr. David Kriegman, and with Kevin Barnes. Our website allows users to upload their pictures and apply makeup to them virtually. We also license our technology to many cosmetic brand sites (Estee Lauder, Revlon, Smashbox, Aveda), media giants (People, InStyle, iVillage, Bravo) and retailers (Marks & Spencer, Topshop).
TAAZ set out to change the way cosmetic products were sold online. Going to a makeup counter allows women to try cosmetic products before buying them. We brought this real-world experience to the online world. Along the way, we discovered that creative expression and experimentation were at the core of the makeup selection and usage experiences, and we therefore built tools that entertained women and helped them express their creativity.
Sight Commerce was launched in 2013, based on the premise that compelling visual imagery can significantly improve conversion and sales for online retailers. Using sophisticated computer vision and machine learning algorithms, Sight Commerce overlays and showcase products (cosmetics, clothing, etc.) on models without needing to shoot photographs of each model-plus-product combination. Costumers include Lancôme, Estée Lauder, Bloomingdales, and Macy's.
Unlike the consumer-facing products we built at TAAZ, our products at Sight Commerce are built for retailers and cosmetic brands to help them increase shop-to-sales conversions via the use of compelling visual imagery. We generate hundreds of millions of images each month to help retailers inspire consumers, understand consumers' shopping preferences, and increase sales by building confidence in the consumers' purchase decisions.
What was your motivation for creating, and what is your aspiration for, the Learn OpenCV website?
Learn OpenCV is a personal side project I started in late Jan 2015; I publish design examples and tutorials for those interested in learning how to use the OpenCV library. Learn OpenCV was born purely out of my desire to help beginners in the crafts of computer vision and machine learning. When my son was learning how to talk, I realized the powerful potential of learning via examples. Imagine trying to teach the rules of grammar first, only showing examples later: a child would never learn to communicate!
In my blog, I usually present a compelling example that tickles the imagination. I follow it by showing the reader how to reproduce my results. Finally, I point them to the theory behind the example. In the long run, I hope to teach people the art of learning quickly, along with the joy and satisfaction of outstanding craftsmanship.
What thoughts do you have about the Embedded Vision Alliance and about its belief that computer vision has now become a mainstream, broadly adopted technology?
I think that the Embedded Vision Alliance is a great initiative, and that computer vision has definitely become mainstream. A very good sign of the broadening adoption of practical computer vision technologies is the fact that programmers with no background in computer vision or machine learning are using libraries like OpenCV and Caffe in order to build interesting software.
What other thoughts do you have that you'd like to share with Alliance website visitors?
Great progress is now being made in the fields of computer vision and machine learning. Join hands and build something wonderful!