VisionBernie - Bernhard Egger
@visionbernie.bsky.social
Prof. @UniFAU. We @CogCoVi study, how humans and computers can perceive the visual world.
(he/him)
0 1 0
0 0 1
1 1 1
https://eggerbernhard.ch
(he/him)
0 1 0
0 0 1
1 1 1
https://eggerbernhard.ch
Heute ist ein schöner Artikel zu unserer AR-Erweiterung in den Erlanger Nachrichten. Vielen Dank an Robin Pfeiffer für die akribische Recherche!
Zum diesen Link hier auf eurem Mobiltelefon öffnen und damit die Bilder in der Cauerstrasse 11 in Erlangen "scannen"
eggerbernhard.ch/AIARt/
Zum diesen Link hier auf eurem Mobiltelefon öffnen und damit die Bilder in der Cauerstrasse 11 in Erlangen "scannen"
eggerbernhard.ch/AIARt/
November 10, 2025 at 8:15 PM
Heute ist ein schöner Artikel zu unserer AR-Erweiterung in den Erlanger Nachrichten. Vielen Dank an Robin Pfeiffer für die akribische Recherche!
Zum diesen Link hier auf eurem Mobiltelefon öffnen und damit die Bilder in der Cauerstrasse 11 in Erlangen "scannen"
eggerbernhard.ch/AIARt/
Zum diesen Link hier auf eurem Mobiltelefon öffnen und damit die Bilder in der Cauerstrasse 11 in Erlangen "scannen"
eggerbernhard.ch/AIARt/
I promised it so often! Glad that we finally made it happen: kudos to @andreead-a.bsky.social , @mert-o.bsky.social Timotei Ardelean, Nilofaar Jorkash
Takuma Nishimura, Julian Stahl, Berke Uysal, Daniel Zieger who are the makers of all the fascinating videos
Quote in Picture: Lee Sedol
Takuma Nishimura, Julian Stahl, Berke Uysal, Daniel Zieger who are the makers of all the fascinating videos
Quote in Picture: Lee Sedol
October 26, 2025 at 4:29 PM
I promised it so often! Glad that we finally made it happen: kudos to @andreead-a.bsky.social , @mert-o.bsky.social Timotei Ardelean, Nilofaar Jorkash
Takuma Nishimura, Julian Stahl, Berke Uysal, Daniel Zieger who are the makers of all the fascinating videos
Quote in Picture: Lee Sedol
Takuma Nishimura, Julian Stahl, Berke Uysal, Daniel Zieger who are the makers of all the fascinating videos
Quote in Picture: Lee Sedol
Wir sind dabei bei der Langen Nacht der Wissenschaften
Im Kinderprogramm dürft ihr ein, gemeinsam mit PsychologInnen entwickeltes, Spiel erleben nacht-der-wissenschaften.de/2025/die-mac...
Abends präsentieren wir die Augmented Reality Erweiterung unseres KI-Experimentes. Animierte Kunst! (1/2)
Im Kinderprogramm dürft ihr ein, gemeinsam mit PsychologInnen entwickeltes, Spiel erleben nacht-der-wissenschaften.de/2025/die-mac...
Abends präsentieren wir die Augmented Reality Erweiterung unseres KI-Experimentes. Animierte Kunst! (1/2)
October 24, 2025 at 6:31 AM
Wir sind dabei bei der Langen Nacht der Wissenschaften
Im Kinderprogramm dürft ihr ein, gemeinsam mit PsychologInnen entwickeltes, Spiel erleben nacht-der-wissenschaften.de/2025/die-mac...
Abends präsentieren wir die Augmented Reality Erweiterung unseres KI-Experimentes. Animierte Kunst! (1/2)
Im Kinderprogramm dürft ihr ein, gemeinsam mit PsychologInnen entwickeltes, Spiel erleben nacht-der-wissenschaften.de/2025/die-mac...
Abends präsentieren wir die Augmented Reality Erweiterung unseres KI-Experimentes. Animierte Kunst! (1/2)
You did publish this year at @cvprconference.bsky.social @eccv.bsky.social @acmsiggraph.bsky.social @ieeevis.bsky.social or a similar?
VMV in Erlangen offers a nectar track where we invite you to present your already published work in a nectar track
🗓️ Submission Deadline: August 15th, 2025
VMV in Erlangen offers a nectar track where we invite you to present your already published work in a nectar track
🗓️ Submission Deadline: August 15th, 2025
August 1, 2025 at 6:19 AM
You did publish this year at @cvprconference.bsky.social @eccv.bsky.social @acmsiggraph.bsky.social @ieeevis.bsky.social or a similar?
VMV in Erlangen offers a nectar track where we invite you to present your already published work in a nectar track
🗓️ Submission Deadline: August 15th, 2025
VMV in Erlangen offers a nectar track where we invite you to present your already published work in a nectar track
🗓️ Submission Deadline: August 15th, 2025
Congratulations, to Saleeta Qadir for winning the worldwide best illusion of the year contest - very proud of her!
Saleeta is just finishing her masters in AI @fau.de and is looking for exciting roles or research opportunities -hire her (or provide me with funding to do it myself)!
Saleeta is just finishing her masters in AI @fau.de and is looking for exciting roles or research opportunities -hire her (or provide me with funding to do it myself)!
July 5, 2025 at 7:48 PM
Congratulations, to Saleeta Qadir for winning the worldwide best illusion of the year contest - very proud of her!
Saleeta is just finishing her masters in AI @fau.de and is looking for exciting roles or research opportunities -hire her (or provide me with funding to do it myself)!
Saleeta is just finishing her masters in AI @fau.de and is looking for exciting roles or research opportunities -hire her (or provide me with funding to do it myself)!
Today is your last chance to vote:
vote.illusionoftheyear.com
vote.illusionoftheyear.com
June 27, 2025 at 6:43 AM
Today is your last chance to vote:
vote.illusionoftheyear.com
vote.illusionoftheyear.com
We proudly present the static spin illusion - a finalist of the Best Illusion of The Year Contest. You perceive a rotation? There is none! Saleeta Qadir student at @fau.de built this illusion using computer vision and image processing - please vote for her:
vote.illusionoftheyear.com
vote.illusionoftheyear.com
June 23, 2025 at 5:52 PM
We proudly present the static spin illusion - a finalist of the Best Illusion of The Year Contest. You perceive a rotation? There is none! Saleeta Qadir student at @fau.de built this illusion using computer vision and image processing - please vote for her:
vote.illusionoftheyear.com
vote.illusionoftheyear.com
Great new world!
The Bergkirchweih in Erlangen starts on Thursday June 6th...
Except there is no Thursday June 6th this year
The Bergkirchweih in Erlangen starts on Thursday June 6th...
Except there is no Thursday June 6th this year
May 18, 2025 at 2:00 PM
Great new world!
The Bergkirchweih in Erlangen starts on Thursday June 6th...
Except there is no Thursday June 6th this year
The Bergkirchweih in Erlangen starts on Thursday June 6th...
Except there is no Thursday June 6th this year
this sounds very lazy...
you need to add at least one binary!
you need to add at least one binary!
April 2, 2025 at 7:58 AM
this sounds very lazy...
you need to add at least one binary!
you need to add at least one binary!
Finally let's revisit “AMOR: Ambiguous Authorship Order” our 2024 #SIGBOVIK contribution that resolved author ordering problems once and for all! We just implement animated author lists eggerbernhard.ch/AMOR.pdf
(and no, our server is not down, your iPhone is just not ready for Science 2.0)
(and no, our server is not down, your iPhone is just not ready for Science 2.0)
April 1, 2025 at 7:17 PM
Finally let's revisit “AMOR: Ambiguous Authorship Order” our 2024 #SIGBOVIK contribution that resolved author ordering problems once and for all! We just implement animated author lists eggerbernhard.ch/AMOR.pdf
(and no, our server is not down, your iPhone is just not ready for Science 2.0)
(and no, our server is not down, your iPhone is just not ready for Science 2.0)
To validate our method, we study (among others) how zero2hero would have influenced the number of citations if it had been used prior to publication (our time machine was used to travel back in time). Results are overwhelming.
Check out the paper for more - you too deserve to be a hero
4/4
Check out the paper for more - you too deserve to be a hero
4/4
April 1, 2025 at 7:09 PM
To validate our method, we study (among others) how zero2hero would have influenced the number of citations if it had been used prior to publication (our time machine was used to travel back in time). Results are overwhelming.
Check out the paper for more - you too deserve to be a hero
4/4
Check out the paper for more - you too deserve to be a hero
4/4
Given a Latex document, zero2hero automatically extracts embarrassingly simple equations and makes them look extremely complex.
We applied zero2hero to various not-so-successful papers (with < 1M citations) published after 2016. Can you spot yours?
3/4
We applied zero2hero to various not-so-successful papers (with < 1M citations) published after 2016. Can you spot yours?
3/4
April 1, 2025 at 7:09 PM
Given a Latex document, zero2hero automatically extracts embarrassingly simple equations and makes them look extremely complex.
We applied zero2hero to various not-so-successful papers (with < 1M citations) published after 2016. Can you spot yours?
3/4
We applied zero2hero to various not-so-successful papers (with < 1M citations) published after 2016. Can you spot yours?
3/4
zero2hero turns every paper into an award-winning masterpiece - and it is heavily used in our community!
How the hell do we do that? Well, based on the fact that papers solely relying on *simple math* are more likely to be rejected, we simply over-complicate equations.
2/4
How the hell do we do that? Well, based on the fact that papers solely relying on *simple math* are more likely to be rejected, we simply over-complicate equations.
2/4
April 1, 2025 at 7:09 PM
zero2hero turns every paper into an award-winning masterpiece - and it is heavily used in our community!
How the hell do we do that? Well, based on the fact that papers solely relying on *simple math* are more likely to be rejected, we simply over-complicate equations.
2/4
How the hell do we do that? Well, based on the fact that papers solely relying on *simple math* are more likely to be rejected, we simply over-complicate equations.
2/4
Tired of seeing your papers being rejected from @cvprconference.bsky.social ? Let's revisit our 2023 #SIGBOVIK paper: "From Zero to Hero: Convincing with Extremely Complicated Math", led by @maxweiherer.bsky.social
Paper: mweiherer.github.io/zero2hero.pdf
Code: github.com/mweiherer/ze...
1/4
Paper: mweiherer.github.io/zero2hero.pdf
Code: github.com/mweiherer/ze...
1/4
April 1, 2025 at 7:09 PM
Tired of seeing your papers being rejected from @cvprconference.bsky.social ? Let's revisit our 2023 #SIGBOVIK paper: "From Zero to Hero: Convincing with Extremely Complicated Math", led by @maxweiherer.bsky.social
Paper: mweiherer.github.io/zero2hero.pdf
Code: github.com/mweiherer/ze...
1/4
Paper: mweiherer.github.io/zero2hero.pdf
Code: github.com/mweiherer/ze...
1/4
Dazzle patterns were a super smart idea to try to hide range, speed and heading of ships
en.wikipedia.org/wiki/Dazzle_...
THIS IS HOWEVER ASSUMING ONE SINGLE VIEW - the moment we get more views, well...
3/4
en.wikipedia.org/wiki/Dazzle_...
THIS IS HOWEVER ASSUMING ONE SINGLE VIEW - the moment we get more views, well...
3/4
April 1, 2025 at 3:22 PM
Dazzle patterns were a super smart idea to try to hide range, speed and heading of ships
en.wikipedia.org/wiki/Dazzle_...
THIS IS HOWEVER ASSUMING ONE SINGLE VIEW - the moment we get more views, well...
3/4
en.wikipedia.org/wiki/Dazzle_...
THIS IS HOWEVER ASSUMING ONE SINGLE VIEW - the moment we get more views, well...
3/4
With so called Erlkönig cars they are trying to hide the shape of their prototypes. We however show that the reconstruction with this pattern turns out fantastic and much better than without that pattern. With just a few photographs we can recover perfect shape.
2/4
2/4
April 1, 2025 at 3:22 PM
With so called Erlkönig cars they are trying to hide the shape of their prototypes. We however show that the reconstruction with this pattern turns out fantastic and much better than without that pattern. With just a few photographs we can recover perfect shape.
2/4
2/4
Three years after this transformative work we presented #SIGBOVIK, there are still Erlkönig cars on our streets!
"A Free Computer Vision Lesson for Car Manufacturers
or It is Time to Retire the Erlkönig"
eggerbernhard.ch/erlkoenig.pdf
Still proud of this work by @maxweiherer.bsky.social
1/4
"A Free Computer Vision Lesson for Car Manufacturers
or It is Time to Retire the Erlkönig"
eggerbernhard.ch/erlkoenig.pdf
Still proud of this work by @maxweiherer.bsky.social
1/4
April 1, 2025 at 3:22 PM
Three years after this transformative work we presented #SIGBOVIK, there are still Erlkönig cars on our streets!
"A Free Computer Vision Lesson for Car Manufacturers
or It is Time to Retire the Erlkönig"
eggerbernhard.ch/erlkoenig.pdf
Still proud of this work by @maxweiherer.bsky.social
1/4
"A Free Computer Vision Lesson for Car Manufacturers
or It is Time to Retire the Erlkönig"
eggerbernhard.ch/erlkoenig.pdf
Still proud of this work by @maxweiherer.bsky.social
1/4
"ACTION: A Catchy Title Is all yOu Need!"
eggerbernhard.ch/action.pdf
another pearl from #SIGBOVIK 2022
Reviewer 2 raised concerns that our title actually proves that it is not only about the title but also an acronym. 3 years later we still refuse to thank the reviewer for their feedback.
eggerbernhard.ch/action.pdf
another pearl from #SIGBOVIK 2022
Reviewer 2 raised concerns that our title actually proves that it is not only about the title but also an acronym. 3 years later we still refuse to thank the reviewer for their feedback.
April 1, 2025 at 3:15 PM
"ACTION: A Catchy Title Is all yOu Need!"
eggerbernhard.ch/action.pdf
another pearl from #SIGBOVIK 2022
Reviewer 2 raised concerns that our title actually proves that it is not only about the title but also an acronym. 3 years later we still refuse to thank the reviewer for their feedback.
eggerbernhard.ch/action.pdf
another pearl from #SIGBOVIK 2022
Reviewer 2 raised concerns that our title actually proves that it is not only about the title but also an acronym. 3 years later we still refuse to thank the reviewer for their feedback.
This work on "(Un)helper functions" was probably the most unhelpful project I was ever involved in - thanks for nothing @thiskevinsmith.bsky.social !
eggerbernhard.ch/unhelper.pdf
This is unhelpful and of no interest to anyone who wants to learn to code, a complete waste of time
#SIGBOVIK
eggerbernhard.ch/unhelper.pdf
This is unhelpful and of no interest to anyone who wants to learn to code, a complete waste of time
#SIGBOVIK
April 1, 2025 at 11:07 AM
This work on "(Un)helper functions" was probably the most unhelpful project I was ever involved in - thanks for nothing @thiskevinsmith.bsky.social !
eggerbernhard.ch/unhelper.pdf
This is unhelpful and of no interest to anyone who wants to learn to code, a complete waste of time
#SIGBOVIK
eggerbernhard.ch/unhelper.pdf
This is unhelpful and of no interest to anyone who wants to learn to code, a complete waste of time
#SIGBOVIK
Following industry standards, we are holding the code and model back from the public to prevent its use by malicious actors.
Since our method is image-based, it generalizes to real world graphs: comment with a 256x256 image and we will add error bars for free!
5/5
Since our method is image-based, it generalizes to real world graphs: comment with a 256x256 image and we will add error bars for free!
5/5
April 1, 2025 at 10:23 AM
Following industry standards, we are holding the code and model back from the public to prevent its use by malicious actors.
Since our method is image-based, it generalizes to real world graphs: comment with a 256x256 image and we will add error bars for free!
5/5
Since our method is image-based, it generalizes to real world graphs: comment with a 256x256 image and we will add error bars for free!
5/5
It is unclear to us how we could have made it more clear that @neurobongo.bsky.social is not an author of this work.
Last but not least - @garymarcus.bsky.social I hope you remember us from twitter/X and we would be thrilled if we can replicate this.
4/5
Last but not least - @garymarcus.bsky.social I hope you remember us from twitter/X and we would be thrilled if we can replicate this.
4/5
April 1, 2025 at 10:23 AM
It is unclear to us how we could have made it more clear that @neurobongo.bsky.social is not an author of this work.
Last but not least - @garymarcus.bsky.social I hope you remember us from twitter/X and we would be thrilled if we can replicate this.
4/5
Last but not least - @garymarcus.bsky.social I hope you remember us from twitter/X and we would be thrilled if we can replicate this.
4/5
Error bars derived in this work are based on n=10'000 datapoints and are much more trustworthy than any other error bar estimation technique.
Our work is also first to derive hyperparameters solely based on MC Hammer album sales (shameless plug)
3/5
Our work is also first to derive hyperparameters solely based on MC Hammer album sales (shameless plug)
3/5
April 1, 2025 at 10:23 AM
Error bars derived in this work are based on n=10'000 datapoints and are much more trustworthy than any other error bar estimation technique.
Our work is also first to derive hyperparameters solely based on MC Hammer album sales (shameless plug)
3/5
Our work is also first to derive hyperparameters solely based on MC Hammer album sales (shameless plug)
3/5
Our idea is simple, yet powerful.
We propose a fully data-driven approach to derive error bars - exactly what you would expect from error bars.
We also achieve superhuman performance, since those error bars are generated faster than by humans.
2/5
We propose a fully data-driven approach to derive error bars - exactly what you would expect from error bars.
We also achieve superhuman performance, since those error bars are generated faster than by humans.
2/5
April 1, 2025 at 10:23 AM
Our idea is simple, yet powerful.
We propose a fully data-driven approach to derive error bars - exactly what you would expect from error bars.
We also achieve superhuman performance, since those error bars are generated faster than by humans.
2/5
We propose a fully data-driven approach to derive error bars - exactly what you would expect from error bars.
We also achieve superhuman performance, since those error bars are generated faster than by humans.
2/5
Forget ChatGPT! This will revolutionize science!
openCHEAT: Computationally Helped Error bar Approximation Tool - Kickstarting Science 4.0
eggerbernhard.ch/SIGBOVIK_ope...
Plenty of happy users since 2021!
1/5
openCHEAT: Computationally Helped Error bar Approximation Tool - Kickstarting Science 4.0
eggerbernhard.ch/SIGBOVIK_ope...
Plenty of happy users since 2021!
1/5
April 1, 2025 at 10:23 AM
Forget ChatGPT! This will revolutionize science!
openCHEAT: Computationally Helped Error bar Approximation Tool - Kickstarting Science 4.0
eggerbernhard.ch/SIGBOVIK_ope...
Plenty of happy users since 2021!
1/5
openCHEAT: Computationally Helped Error bar Approximation Tool - Kickstarting Science 4.0
eggerbernhard.ch/SIGBOVIK_ope...
Plenty of happy users since 2021!
1/5
"self driving is about two scalars" is what @jonbarron.bsky.social claimed in his keynote @3dvconf.bsky.social and he could not be more wrong!
And this besides our impactful publication from 2020 on level 6 autonomous driving!
sigbovik.org/2020/proceed...
When will industry finally listen to us?
And this besides our impactful publication from 2020 on level 6 autonomous driving!
sigbovik.org/2020/proceed...
When will industry finally listen to us?
April 1, 2025 at 8:49 AM
"self driving is about two scalars" is what @jonbarron.bsky.social claimed in his keynote @3dvconf.bsky.social and he could not be more wrong!
And this besides our impactful publication from 2020 on level 6 autonomous driving!
sigbovik.org/2020/proceed...
When will industry finally listen to us?
And this besides our impactful publication from 2020 on level 6 autonomous driving!
sigbovik.org/2020/proceed...
When will industry finally listen to us?