Winning HackZurich 2018

by Heiki Riesenkampf, 18.10.2018
Team “deepbuster” receiving the main prize from the organizers Jonathan Isenring (left) and Rasmus Rothe (second from right) (photo credit: HackZurich)

Every great journey starts in a bar and so did ours. The first time we met was around 6 months ago, on a cold winter night in BQM – ETH Zurich’s student bar. We had attended a lecture about entrepreneurship held by Lakestar Venture Capitalists. The organizers of the event offered to continue the evening at the BQM to have a beer and chat about their next big startup ideas. As a new student in Zurich, I felt this was a great opportunity for me to meet and get to know more of the local ETH crowd.

As it turned out, only a handful of others took Lakestar up on their offer. Igor Susmelj and I just happened to be among them. Even though I hadn’t met Igor before this evening, he seemed like a cool fellow and I decided to introduce myself. After some brief small talk we realized that we shared a lot more in common than we had ever expected. We were both working on our master thesis and, as it just so happens, we were to about to be working in the same lab. To put the cherry on the cake, we discovered that we even were assigned to the same supervisor, Prof. Luc van Gool. More specifically, we were both doing research on Generative Adversarial Neural Networks (GAN) in the Computer Vision Lab at ETH Zurich.

At that time, Igor was trying to delete cars from streets and I was trying to make vacation pictures out of nothing! In addition to our common research interests, we also shared a passion for entrepreneurship and the start-up world. Eventually, looking to the future and what the world might hold for us, it made absolute sense for us to pursue a path toward starting our own company. Having discovered an incredible amount of shared goals, ambitions, and passions our paths seemed to align much sooner than we thought. Working side-by-side on our research to prepare the ground for some promising future ventures, as a duo we got started. It did not take long to realize that our common skillset might come in handy in a competition too.
Team "deepbuster" pitching at HackZurich 2018 (photo credit: @alanbilalloicsahin)

Registration to HackZurich 2018? Done!

Until the hackathon, our thesis demanded all of our attention and had us working day and night, all week long, and in sun or snow – and the attention was worth it. Our research took us into some uncharted territory and the excitement pushed us even more. Two minds, like ours, were baffled by the potential of the technology we were developing in the lab and we harbored the idea of a further innovating in this field. Any advance we could bring to this research would also bring us one-step closer to a promising startup. The work went on, the motivation rose, and the amount of sleep fell. A couple of days turned into a couple of weeks, and then into months.
Fast forward to the HackZurich 2018 Kickoff evening. Bringing our newly acquired Swiss punctuality to play, we met early as one last challenge was in our way - find more team members. We reached out to a couple of people and managed to convince two excited designers to join our team - Tris-Denny Leyh-Bannurah and Nicolas Zahnd. We needed Tris and Nicolas in order to build the most beautiful fake video detector of them all. Tris is from Germany and brought a fresh business expertise from HSG – University of St. Gallen in Switzerland. Nicolas adds a Swiss touch and beauty to our design prototype.

Without reservation, the next two days of the hackathon were full of hard work. As any learning process starts, this one kicked off with a round of massive downloads of countless “DeepFake” videos that we could find in the four corners of the Internet. Stick all that data into a state-of-the-art artificial brain and it will learn to identify the difference between tampered and original footage. We added some additional algorithms, here and another model there. Then we added more flesh to our meager solution. The project concluded with a demonstration interface on a sexy website and curious minds could take the system for a testdrive.

After two sleepless nights, barrels of energy drinks, towers of pizza, jugs of coffee, and, of course, a teeny tiny bit of coding, we managed to release our detection tool just shy of the final submission deadline. On the final day, the hackathon organizers selected our team, “deepbuster” to present our work on the main stage. Our whole team managed to descend from the mountain of excitement, just in time to go on stage and make a great presentation of our vision and work.
The glory pitch is available here
Half an hour later, the judges’ hammers had fell and they declared our project as the winner of the competition. We did it! But, the most precious prize we ended up taking home, was simply the bond that we formed between the members of this “dream team.” And if you thought our journey would end back in BQM ordering drinks for 5k, then unfortunately, I have to disappoint you, because it is not the case.

Igor and I have now finished our master thesis research and we decided to keep working on the project started during the hackathon. We are going to try to turn it into a company. The vision about starting a company after our studies now seemed so much more clear. Empowering people to make sure that the videos they see online are genuine is a super exciting problem and we are very motivated to solve it.
So what is the moral of the story? Get out of your comfort zone and approach that random person in the bar,
because you never know what might come out of it.
Want to learn more about the “deepbusters” or help us fight fake media?
Get in touch with us!
Send an email

About the author

Born and raised in Estonia, Heiki Riesenkampf enjoyed an idyllic childhood growing up in the countryside. After spending some time on a high school exchange program in the US he graduated high school in Tallinn and started university at Imperial College London in Computer Science. Having always being entrepreneurial-minded he started his first company at the end of high school and worked on some startup projects while living in London. After getting some working experience at Credit Suisse and Google he decided to finish his master studies as an exchange student and come to ETH Zurich for his final year. Here he finished his master thesis at the Computer Vision Lab with Luc van Gool working on generating photo realistic images with adversarial networks. Now, after graduation, he is looking forward to starting a new company in the computer vision field called "Mirage" which is specialised on fighting against fake media. When not working, he really enjoys spending time on the lake or hiking in the mountains in Southern Switzerland.