Excire launches AI-powered image management tool, Excire Foto

Gary Pageau of the Dead Pixels Society talks with Erhardt Barth, co-founder of Excire Inc., about the company's new desktop-based image management tool, Excire Foto. Excire Foto is a powerful app designed for easy photo management and quick content-based browsing. As a local program, it doesn't require cloud computing and therefore has built-in privacy. Pageau and Barth discuss the company's background, the technology, the need for intelligent image management, and the role of privacy in AI.

Excire Inc. is an American corporation dedicated to bringing useful AI technology to photographers. The Excire technology is developed by the Pattern Recognition Company (PRC), a German AI firm started in 2005 as a spin-off of the University of Lübeck.

Professor Barth is co-founder of the Pattern Recognition Company and Excire Inc. and CEO of PRC. He has a PhD in electrical engineering from the Technical University of Munich, did post-doctoral work at NASA Ames, and is an internationally recognized expert in Computer Vision and Machine Learning. He founded and co-founded several companies including the gestigon GmbH, now a Valeo brand.

Transcript

Intro - Erin Manning (2s):
Welcome to the Dead Pixels Society podcast. The photo imaging industry's leading news source. Here's your host, Gary Pageau

Gary Pageau (11s):
The Dead Pixels Society podcast is brought to you by Photo Finale, Mediaclip, Advertek Inc. GotPhoto, EyeQ, Photobook AI, The IPI member network and MailPix. Hello, and welcome again to the Dead Pixels Society podcast. Today we're joined by Erhardt Barth, the CEO and cofounder of Excire, a company out of Germany that is making waves in the world of artificial intelligence.

Gary Pageau (42s):
Hello, Erhardt. Welcome.

Erhardt Barth (45s):
Hello. Thanks for having me.

Gary Pageau (47s):
For those people who aren't familiar with your company, can you tell a little bit about the story of the background? It's actually been around over 10 years, is that correct?

Erhardt Barth (59s):
It's 15 years now. And as the name says the Pattern Recognition Company, so we are doing AI and we both found us have a background, you know, in the area of neural networks and AI, which is becoming popular now, but we have been doing this for ages.

Gary Pageau (1m 24s):
So it's a, it's a spinoff of a company called pattern recognition company.

Erhardt Barth (1m 29s):
That's correct.

Gary Pageau (1m 31s):
Can you talk a little bit about the technology? What do you mean by neural network?

Erhardt Barth (1m 39s):
I mean, these, these acronyms that are inspired to a certain degree by how the brain works, right? So many know about deep learning now. So these large deep networks that you can train, if you have large amounts of data, basically, that's what we did. We collected huge amount of data, you know, with different content and then you can use these deep networks to train them, to recognize what is in the image.

Erhardt Barth (2m 11s):
Right? So that's the idea.

Gary Pageau (2m 14s):
If you're going back to the mid-2000, you're talking probably possibly millions, maybe even billions of images accrued over that time. So that's a, that's a huge knowledge base to build from.

Erhardt Barth (2m 31s):
Yeah. I mean the, the increasing number of images is good and bad, right? So it's good for us if we can use them to train the AI that because you're, you know, it's, you get flooded by all these images. And this is actually why I started to work on this because I've been a photographer and I'm a tour photographer for a long time. So when I, as a, as student, I learned my money with photography, right.

Erhardt Barth (3m 3s):
And then, you know, when I switched to digital photography and accumulated more and more images, I realized that they kind of, you know, become worthless if you can't find the images that you are looking for. Right. So, so then I, then we decided to use the knowledge that we had because before we did the exile, we had, you know, mainly projects in machine vision and medical applications, you know, doing diagnose diagnosis and stuff like that.

Erhardt Barth (3m 36s):
And then we decided to do something for photography was kind of my personal project because I needed it by myself.

Gary Pageau (3m 47s):
Sure. Yeah. Well, that was one of things I see on your website here, where they talk about the history of the company that, you know, you had an image collection of more than a hundred thousand images, then you realized, Oh my gosh, you know, I can't find these things.

Erhardt Barth (3m 59s):
Yeah. I think I forgot that we had that on the page.

Gary Pageau (4m 4s):
Yeah, a little bit about why your technology is better suited for image management than some of the other solutions that are out there. Cause you, cause your first products were actually plugins to other products. Is that correct?

Erhardt Barth (4m 22s):
We started by doing a plugin for Lightroom. We thought that would be an easy start, but turned out to be quite complex in the end. And because we thought, you know, many people are using Lightroom and would be useful to have this plugin.

Gary Pageau (4m 42s):
What did the plugin actually do in terms of what was it tagging or what was the, was it, was it doing image recognition within the catalog?

Erhardt Barth (4m 52s):
Yeah, it was within the catalog. Yeah. And then you could also use it to tag the images and use them somewhere else, but mainly you could use it within Lightroom in the catalog and it's not only tagging. It's also, I think the really useful thing is if you combine the different features, right. If you, if you tag and search by keywords, but then you can also search for similarities. And then we further develop that too. So you can combine all different criteria. For example, if eyes are open, if people are smiling, you can combine it with under.

Gary Pageau (5m 27s):
Yeah. So, so theoretically you could search off for all of your pictures of smiling people, outdoors wearing a certain, like a wedding dress, right. If you're a girl photographer you could find. Right.

Erhardt Barth (5m 41s):
Right. And that actually, I think that generates the value that you can combine these different searches and then yeah.

Gary Pageau (5m 48s):
Yeah. Cause I mean, I mean, it seems to me that, you know, you know, tagging and sort of that sort of simple cataloguing has been around for some time, like where it's, you know, this is a portrait outdoor shot. It doesn't really get very deep. And like you said, it's almost like creating a script almost. Right. I want to find this type of picture..

Erhardt Barth (6m 10s):
But there is one more difference if you're thinking of a, you know, Adobe and Google and Apple, so they are all doing this in the cloud. We just find it easier if you can, you know, use your servers in the cloud to analyze. But we managed to build, let's say much, much more efficient networks that you can then install on the people's computers and then you can perform all the analysis on the computer.

Erhardt Barth (6m 40s):
So you don't need, you don't need to upload things to the cloud. Right. In order to get your images analyzed as a number of advantages. I think now it's like privacy, but also it takes a long time. Right.

Gary Pageau (6m 57s):
Yeah. I was going to bring up the privacy aspect. Cause I think that's actually something that's pretty important these days, especially for like professional photographers, people doing work that, you know, they want maintain the, you know, the, the use of, you know, sending images across the cloud to be analyzed, I think is not really something that people, a lot of people are going to be open to even, I mean, even on the amateur side, right. I mean, we're getting people concerned about AI recognition happening in the cloud.

Gary Pageau (7m 31s):
There's been, there's been discussion, you know, some of the online recognition providers like Amazon or, you know, they're kind of backing off on some of their offerings because of some of the concerns on privacy. I think that's a great differentiator with your product. So are you still offering the, the plugin tool before we start talking about your newer products? Are you still offering the plugin to the, the Lightoo?

Erhardt Barth (7m 55s):
Sure. Yeah. It's still around. So I mean I'm in the plugin and in the new standalone we are using the same engine. Right. So we're going to further develop this and we'll so both were benefited by the plugin and the standalone. Right. Cause Excire Foto.

Gary Pageau (8m 14s):
Yeah. So, so let's talk a little bit of a et cetera photo. It is, it is a standalone product it's now available in the U S it has been around overseas for a while, right?

Erhardt Barth (8m 26s):
Yeah. But only a few weeks earlier I started. Yeah. Yup.

Gary Pageau (8m 30s):
So now you have a, it's a standalone product. Can you tell us a little bit about that? Some of the requirements required for that kind of the ideal person, the ideal environment it's suited for, for example,

Erhardt Barth (8m 45s):
I mean, I don't think that there are any major constraints, right? So of course our customers so far let's say professionals and maybe the, how do you call them the amateurs who are more enthusiasts and they would typically have, you know, a larger collection and maybe they have their own rates for backups and stuff, because, you know, regarding the privacy issue that we have before, I think it's much better to have your own cloud.

Erhardt Barth (9m 22s):
Right. And it's easy to create your own cloud with a rate system. And of course you need some kind of backup, but why, why not have your own, right. Yeah. So, Excire Foto is something we did for a number of reasons. One is because many people asked us people who are not using Lightroom, but then also the other reason is that we, we could design it more flexible.

Gary Pageau (9m 58s):
Sure, but you're not relying on lay on the Lightroom interface and the light.

Erhardt Barth (10m 2s):
Right, right, right. And we could really design it for finding images. Right. And taking images. Of course. Yeah.

Gary Pageau (10m 11s):
So once you've got your images, you know, organized, and you've got some of these collections built, you know, what are some of the things you can do with them? Right. I mean, you can obviously, can you, you know, share them some other things, what can you do? I mean,

Erhardt Barth (10m 33s):
Finding images is something you need all the time. And so I realized that in the beginning, when I was working on digital images, right, you spend a lot of time, you know, developing and editing, but then you, sometime you should get a routine in developing and issue, you know, maybe sometimes avoid developing, thinking more about taking a picture. Then you spend most of the time searching for images of, for example, you want to do a book, then you spend a lot of time searching for the right images. You want to share some images or you want to, you know, so the, so culling and searching images, you know, it takes a lot of time and that's where we, we want to create a value.

Erhardt Barth (11m 17s):
Right. So you can quickly call your images and the ones you're looking for. Right. So that's the, the idea.

Gary Pageau (11m 26s):
So, but, but I mean like one of the things I know that you can do is you can obviously post collections too. You know, if you want to share them, you can do Dropbox or Google or things like that. Being also to support the keywords and some of the other data. Right. How would somebody use that? Okay.

Erhardt Barth (11m 42s):
I mean, you can analyze, you can use our exact photo to analyze the images and then you get all the key words or the text, then you can export them right into the image files or the site kind of fires. And then you can use them in Lightroom or capture one or whatever. Right. So we have many customers who asked us about capture one and now they can, no, they can import the images to examine photos of them, call them a keyword, them, analyze them and then take them over to capture one.

Erhardt Barth (12m 22s):
And,

Gary Pageau (12m 23s):
And so, so, so they could, for example, like you said, in that, in that workflow experience, they could, you know, capture images of, of an event, use Excire Foto to gather the best one and then use that information in there or are there, and there are more advanced editing application, like capture one or whatever, and it will actually write the data into the image file. So you're not losing that.

Erhardt Barth (12m 54s):
Right. Right. Yep. Yep. I mean, you could, I mean, we have a sketch different workflows, right? Of course we would prefer to workflow that exact photo photos, your main program, you can quickly find the images, analyze them. And then if you need to do some editing, you can pick the, you know, a raw editing software that is best suited for that particular image. Right. So, because if you should Fujis then maybe you want to use one editor.

Erhardt Barth (13m 25s):
If you should, another format, you might want to use another one. Right. So, but then you can keep excited for that as your main program and then use different programs to edit or, but you could also in our stick to Europe. So if you are used to Lightroom and capture one, you can stick to those. But if there's intermediate step of keywording all the images and cutting them and then go into your application. Right.

Gary Pageau (13m 52s):
So I think that's one of the things you hit on a real need for that people have, because as ever not everyone has even a, you know, a wash with all these images permanently in the smart phone space. How is this useful to someone who's, let's say, cause more and more even advanced San Francisco pros are integrating smartphones into their workflow. How, how, how does exciter photo just treat that as another, as a separate camera? Just, just another input source, just another cameras or anything special there.

Erhardt Barth (14m 24s):
Yeah. Just another camera. Right. But you know, the thing is that, for example, on your phone, you're not really able to determine whether it's really Sharm and stuff like that. And so that's something we are going to invest a lot of development in the future that we, we give some tools that you can, for example, you can find out whether the eyes are sharp and stuff like that without zooming in.

Erhardt Barth (14m 54s):
And so that's, I think where you can profit also with fuse shoot on, on phones that you could then get a rating of, you know, sharpness and maybe overall quality and so on. So you can pick the pick the best images and move on with those only. Right. So it could even,

Gary Pageau (15m 17s):
For example, a lot of smartphone photographers use burst mode, for example, something like that, where they're taking. Right. So you could select the best one out of that group. Yeah.

Erhardt Barth (15m 29s):
So we actually, what we were going to release that pretty soon. So we will have a cutting module where we automatically determine, which is the same, how would you say the same motif or the same shot? Yeah. But then you might have like five, 10, 20 images of the same shot. And then we will order them in, you know, regarding to sharpness or whether the eyes are open or closed or you could pick, you know, if you want to pick the one where people smile, you can just tick that and then you will select those where people's minds and things like that.

Erhardt Barth (16m 12s):
So that will make it very easy to, okay. If you have all these bursts shards to pick the best one, because that's where you spend a lot of time when you're looking at tenure, one of find out which one is the best of these 10. It takes takes a while.

Gary Pageau (16m 27s):
Yeah. Especially when you're talking, I mean, you know, if you're, you gotta, you're trying to compare two images and they may be essentially similar, right. To have a little extra AI help in the background, right. To help with that, to maybe find that little, little sharper or the little one with little better contrast when you have two images, which may or may not be.

Erhardt Barth (16m 51s):
Yeah. And also higher level of features, like whether the eyes are closed or the, so these are,

Gary Pageau (16m 59s):
So right now it doesn't appear like there's much really in terms of editing features. Right. Is that really just something you're kind of saying, Hey, listen, we're an organizing tool. We're not a full-blown editing feature program.

Erhardt Barth (17m 16s):
Yeah. So that was our conclusion while it doesn't make sense to compete with the editing tools which are around. Right. Because everybody has his kind of favorite editor and in some cases, for example, some I'm using Fuji cameras and then there's also, you can edit them with the Fuji tools which is pretty useful. So you can pick the editing tools that, and then we have, we provide an interface so that you can open the image and edited and then yeah.

Erhardt Barth (17m 54s):
But we will, of course give some basic editing tools with in the future. Like just, you know, we have to think about that. What would be like a minimum set of editing features? I'm sure. Like, you know, a typical crop and cropping and stuff like that. Yeah, yeah. Yeah. Sounds like a very, very exciting product. Definitely something, I think a lot of people will want to experiment with you. Haven't how can somebody get a evaluation copy or purchase it?

Erhardt Barth (18m 24s):
Well, that's pretty easy to just go to site exotic.com and you can download a test version, which you can use for 30 days. And if you like it and it's, you know, just Pape on and you're done, you're all about the ease of use as that. Right? Yeah. Yeah. So what's been the response since you've launched the, the desktop version, it's been out a couple of weeks now, right?

Erhardt Barth (18m 56s):
Yeah. I mean, in, in Europe, it's it's good. It's very good actually. So we get a lot of interest in, in U S it's kind of slow, but we just started, but we are really excited to see what's going to come. Right. Well, great. Well, listen, thank you for your time, sir. I appreciate you reaching out from across the pond over in Germany and wish you well and look forward to keeping up in the future with more updates.

Erhardt Barth (19m 28s):
Yeah, that would be nice. So yeah. Thanks for having me again, right. Thank you

Outro - Erin Manning (19m 37s):
Thank you for listening to the Dead Pixels Society podcast. For more great stories and sign up for the newsletter at wwwthedeadpixelssociety.com.