I'm Sven. I am a postdoctoral research associate at Indiana University's School of Informatics, Computing, and Engineering, working with David Crandall, Chen Yu, and Linda B. Smith. This website aims to give a brief overview of what I do. For a more in-depth, printable summary, please refer to my CV.
My main research interest is computer vision, i.e. the intersection of computer science, machine learning and artificial intelligence that investigates methods of analyzing and understanding the visual world. My Ph.D. work focused on vision algorithms for first-person (egocentric) cameras that approximate a person's field of view. Motivated by the recent success of deep neural network models in vision, and inspired by many collaborations with developmental psychologists, my current work aims at exploring the interdependency of human learning and machine learning. Can understanding visual learning in toddlers help us improve artificial vision models, and can we use artificial vision models as proxies to help us better understand human vision?
@inproceedings{gaze2018bmvc,
title = {From Coarse Attention to Fine-Grained Gaze: A Two-stage 3D Fully Convolutional Network for Predicting Eye Gaze in First Person Video},
author = {Zehua Zhang and Sven Bambach and David Crandall and Chen Yu},
year = {2018},
booktitle = {British Machine Vision Conference (BMVC)}
}
@InProceedings{Bambach_2017_ICCV_Workshops,
author = {Bambach, Sven and Zhang, Zehua and Crandall, David J. and Yu, Chen},
title = {Exploring Inter-Observer Differences in First-Person Object Views Using Deep Learning Models},
booktitle = {The IEEE International Conference on Computer Vision (ICCV) Workshops},
month = {Oct},
year = {2017}
}
@article{bambachegocentric,
title={An Egocentric Perspective on Active Vision and Visual Object Learning in Toddlers},
author={Bambach, Sven and Crandall, David J and Smith, Linda B and Yu, Chen}
}
@article{bambach2016active,
title={Active Viewing in Toddlers Facilitates Visual Object Learning: An Egocentric Vision Approach},
author={Bambach, Sven and Crandall, David J and Smith, Linda B and Yu, Chen},
booktitle={Proceedings of the Cognitive Science Society},
volume={38},
number={38},
year={2016}
}
@inproceedings{bambach2016objects,
title={Objects in the center: How the infant's body constrains infant scenes},
author={Bambach, Sven and Smith, Linda B and Crandall, David J and Yu, Chen},
booktitle={Development and Learning and Epigenetic Robotics (ICDL-EpiRob), 2016 Joint IEEE International Conference on},
pages={132--137},
year={2016},
organization={IEEE}
}
@InProceedings{Bambach_2015_ICCV,
author = {Bambach, Sven and Lee, Stefan and Crandall, David J. and Yu, Chen},
title = {Lending A Hand: Detecting Hands and Recognizing Activities in Complex Egocentric Interactions},
booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
month = {December},
year = {2015}
}
@inproceedings{bambach2015viewpoint,
title={Viewpoint Integration for Hand-Based Recognition of Social Interactions from a First-Person View},
author={Bambach, Sven and Crandall, David J and Yu, Chen},
booktitle={Proceedings of the 2015 ACM on International Conference on Multimodal Interaction},
pages={351--354},
year={2015},
organization={ACM}
}
@InProceedings{Lee_2014_CVPR_Workshops,
author = {Lee, Stefan and Bambach, Sven and Crandall, David J. and Franchak, John M. and Yu, Chen},
title = {This Hand Is My Hand: A Probabilistic Approach to Hand Disambiguation in Egocentric Video},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2014}
}
@inproceedings{bambach2014detecting,
title={Detecting hands in children's egocentric views to understand embodied attention during social interaction},
author={Bambach, Sven and Franchak, John and Crandall, David and Yu, Chen},
booktitle={Proceedings of the Cognitive Science Society},
volume={36},
number={36},
year={2014}
}
@inproceedings{bambach2013understanding,
title={Understanding embodied visual attention in child-parent interaction},
author={Bambach, Sven and Crandall, David J and Yu, Chen},
booktitle={Development and Learning and Epigenetic Robotics (ICDL), 2013 IEEE Third Joint International Conference on},
pages={1--6},
year={2013},
organization={IEEE}
}
@article{bambach2015survey,
title={A survey on recent advances of computer vision algorithms for egocentric video},
author={Bambach, Sven},
journal={arXiv preprint arXiv:1501.02825},
year={2015}
}
I've been lucky enough to have gained a variety of teaching experiences at IU, working together with many talented people. I've been involved in teaching/designing graduate, undergraduate, and online courses.
In the Fall 2016 semester IU offered CSCI B551 - Elements of Artificial Intelligence as an online course for the first time (in parallel with the residential version of the course). Prof. Crandall and I shared responsibilities for teaching both, with me being the lead-instructor for the online course. Together, we designed and recorded interactive video lectures that were well-received by the students:
- I liked the online videos and lecture style and assignments.
- Timely and substantial comments and replies, easy logistics around the course, good on-line videos.
- The method of teaching and stimulating thinking. A lecture was divided in multiple short videos with assignments in between which helped us understand what we learned and what not.
What did you like most about this course and instructor? Answers are from an anonymous course questionnaire.
During my first two years as a graduate student I helped teaching the following courses:
C211 is the main introductory course to computer science at IU and is attended by hundreds of students each semester. While there is a large team of teaching assistants to handle the logistics, hosting office hours and labs still provides opportunities to make personal impacts:
- Great instructor, always willing to help and give advice. Never talks down to students. Makes sure the students understand.
- Sven is very patient and willing to help his students. He is a good instructor.
- Sven was very helpful, encouraging and very open (meaning there was no hesitation to ask him questions).
- He's a cool dude. Imitates Germans very well.
- Friendly, quick and to the point, but still very clear. Very capable.
- Sven is really knowledgeable about a broad range of topics in computer science and is always eager to share his knowledge with others.
- He was really helpful! The best TA in C211!
What did you like most about this course and/or the TA? Answers are from an anonymous course questionnaire, Fall 12 + Spring 13.
I have also been asked to give various guest lectures in computer vision-related courses both at the graduate and undergraduate level (e.g. CSCI B490 - Introduction to Computer Vision or CSCI B657 - Computer Vision).
I've been working in several academic roles since joining IU. During my undergraduate time and in the time between graduating and going to the U.S., I could also gather experiences in different industrial appointments.
I have a dog, Mooka, who was the "dog of honor" at my wedding.
I biked regularly for a while, and when our computer vision lab received a new GoPro, I "tested" its battery by recording several biking trails in Bloomington. Result: I got exhausted before the battery ran out. Here's the one-hour video... – More recently, I picked up running as well, mostly because I was jealous of my wife after she ran her first half-marathon.
Inspired by playing foosball in-between classes, a friend and I developed a software capable of tracking all key components of a foosball game in real time (30+ FPS). It was a project for a computer vision class that we took in the Spring 2012 semester and a lot of fun to realize. A video of our results can be seen here.
Lastly, I don't just like to work on algorithms that analyze photos, I also love taking them! Although I took a few photography classes as an undergraduate, I am far from being a professional. However, I am a proud owner of a Canon SLR camera, and you can check out a very sparse selection of my photos below:
