Photo management is probably the most disputed space in photo:tech right now with major players like Apple, Google, Microsoft, Yahoo and Amazon, competing for the absolute control of where people store, manage and back up their photos. We sat down with Peter Welinder from Dropbox, a major player in this space, to discuss how computer vision and machine-learning is and will be a major differentiator and success trigger. Peter will be a speaker and judge at the upcoming LDV Vision Summit.
Tell us a little bit about yourself..
I’m currently leading a computer vision and machine learning-oriented
team at Dropbox. We’re solving problems around automatic image
organization in Carousel, our dedicated photos app. Before Dropbox, while in grad school at Caltech, I co-founded a startup called Anchovi Labs. We started out by trying to automate scientific image
annotation, but quickly found out that the market wasn’t ready for that. Instead, we pivoted into image organization for personal photo
How did the sale to Dropbox happen?
We realized pretty quickly that a good file syncing experience would be crucial for a cloud-based photo management app, since users don’t want to abandon the file-based workflows they’re already used to. But we also understood how much work it’d be to build that from scratch. Everyone on the team was already a Dropbox user, and we were extremely impressed by how simple and reliable the Dropbox product was. As we started discussing a potential partnership with Dropbox, it quickly became clear that they had the same vision that we did. One thing lead to another, and we decided to join forces and help start a photos-focused product team at Dropbox.
Does Dropbox store more visuals than text today?
Our users store their most important files in Dropbox. If you look at your own files, you’ll find that a big fraction of them are photos. That’s why we care so much about photos at Dropbox — people need an easy way to save and share their most important memories as much as they do their most important documents.
What is the number one challenge facing Dropbox and photos? What is
the company trying to solve?
We built Carousel to be the home of all your memories. That means we not only want to keep your photos safely backed up, but we also want to provide you with ways to go back and relive your memories and share them with your closest friends and family. We are constantly experimenting with new ways to make the photos experience better, while at the same time ensuring that our solutions are both simple and intuitive.
As an engineering manager, do you have a lot of input on the solutions
being worked on?
Everyone on the Carousel team is passionate about photos, and many are avid photographers. So there are always strong opinions involved in any decision we make, and everyone has input. As an engineering manager, I work closely with designers and product managers to set the vision and prioritize the work to bring the best new features out the door.
Deep learning, object recognition has come a long way since you sold
Anchovi Labs. What do you see today that really excites you ?
Deep learning has become a crucial component for visual recognition, and is already playing a big role in photo-related products out there. For me, it’s also really exciting to see vision technology move from servers to mobile clients. Our phones are packed with sensors and computational power which people can use to capture their memories in richer and more engaging ways. We’ve barely scratched the surface for what’s possible there.
What would you like Dropbox to offer that technology does not yet
allow you to deliver?
Since many of our users store all the photos they’ve ever taken on Dropbox, there are billions of photos that are missing timestamps and GPS data. This makes it so much more difficult to help our users organize and search those photos. One big challenge ahead will be to use computer vision methods to try to infer as much as we can about such content, to make it indexable and searchable based on very little information.
Tell us what is interesting and unique about the LDV Vision Summit?
There aren’t many opportunities for entrepreneurs, companies, students and researchers interested in photography, image processing, and computer vision to get together. The LDV Vision Summit is an amazing opportunity to learn about interesting ideas and trends, and meet people interested in solving similar problems.
Entrepreneurial Computer Vision Challenges? Who should compete and why?
We’re at a unique point in time where computer vision and visual recognition are starting to work really well. It’s far from perfect, and there’s a lot of work ahead, but these technologies have begun to make their way into many consumer products. The Vision Challenges is a great opportunity for students and entrepreneurs to apply what they’ve learned in class and through research to real world problems. People outside of the field are just starting to realize how far it’ll go. This is your chance to inspire them and show what’s possible.
What is your role for the 2015 LDV Vision Summit?
I’ve helped out developing the Vision Challenges and I look forward to being a judge during the summit.
What would be, for you, a sure sign of the summit success?
The summit is a great opportunity for researchers, students,entrepreneurs, companies and investors to get together. A sign of success would be to see some of those connections develop into new jobs, startups, or investments.
[ NDLR : Kaptur is a proud media partner of the LDV Vision Summit and will be bring you privileged coverage of the event and its guests. Stay tuned……]
Photo by Gryphus31
Author: Paul Melcher
Paul Melcher is the founder of Kaptur and Managing Director of Melcher System, a consultancy for visual technology firms. He is an entrepreneur, advisor, and consultant with a rich background in visual tech, content licensing, business strategy, and technology with more than 20 years experience in developing world-renowned photo-based companies with already two successful exits.