Emre Aksan

Emre Aksan

PhD student
Advanced Interactive Technologies Lab, ETH Zürich

E-Mail
eaksan@inf.ethz.ch
Address
Stampfenbachstrasse 48, 8092 Zürich, Switzerland
Room
ETH Zurich, Department of Computer Science, STD, H 27

Biography

I am a PhD student at ETH Zurich, working in the Advanced Interactive Technologies lab with Professor Otmar Hilliges. I received my BSc (2013) and MSc (2015) degrees from Middle East Technical University (METU) in Computer Engineering. Prior to joining ETH, I worked on pattern analysis of functional magnetic resonance imaging at METU.


Research Interests

My research interests lie at the intersection of machine learning, computer vision, and human-computer interaction, with a primary focus on the perception and synthesis of human activities, aiming to digitize humans in various aspects. Technically, I am interested in deriving new machine learning algorithms, particularly generative temporal models to capture human dynamics and generate human-like interactions. My current research focuses on deep generative temporal models with applications in the tasks of 3D motion modeling and prediction, building 3D face avatars, and modeling and synthesis of free-form human actions such as digital representations of drawings and handwritten text.

If you are interested in Semester/Master project in our group, you can check our available projects. If you have your own topic related with my research, you are welcome to contact me.

You can also find me on Twitter, Linkedin, and Google scholar.


Publications

Teaser-Picture

LiP-Flow: Learning Inference-time Priors for Codec Avatars via Normalizing Flows in Latent Space

AuthorsE. Aksan, S. Ma, A. Caliskan, S. Pidhorskyi, A. Richard, S. Wei, J. Saragih, O. Hilliges
In ProceedingsEuropean Conference on Computer Vision (ECCV), 2022
ProjectPDFVideoBibTeX
Teaser-Picture

Reconstructing Action-Conditioned Human-Object Interactions Using Commonsense Knowledge Priors

AuthorsX. Wang*, G. Li*, Y. Kuo, M. Kocabas, E. Aksan, O. Hilliges
In ProceedingsInternational Conference on 3D Vision (3DV), 2022
* These two authors contributed equally to this work.
ProjectPDFVideoBibTeX
Teaser-Picture

D-Grasp: Physically Plausible Dynamic Grasp Synthesis for Hand-Object Interactions

AuthorsS. Christen, M. Kocabas, E. Aksan, J. Hwangbo, J. Song, O. Hilliges
In ProceedingsComputer Vision and Pattern Recognition (CVPR), 2022
ProjectCodePDFVideoBibTeX
Teaser-Picture

A Spatio-temporal Transformer for 3D Human Motion Prediction

AuthorsE. Aksan, M. Kaufmann, P. Cao, O. Hilliges
In ProceedingsInternational Conference on 3D Vision (3DV), 2021
ProjectCodePDFVideoBibTeX

CoSE: Compositional Stroke Embeddings

AuthorsE. Aksan, T. Deselaers, A. Tagliasacchi, O. Hilliges
In ProceedingsAdvances in Neural Information Processing Systems, 2020
ProjectCodePDFBibTeX

Learning Functionally Decomposed Hierarchies for Continuous Control Tasks With Path Planning

AuthorsS. Christen*, L. Jendele*, E. Aksan, O. Hilliges
In IEEE Robotics and Automation Letters (Volume: 6, Issue: 2), 2021
* These two authors contributed equally to this work.

Accepted as Oral Presentation at the Deep RL Workshop at NeurIPS
ProjectCodePDFVideoBibTeX
Teaser-Picture

Convolutional Autoencoders for Human Motion Infilling

AuthorsM. Kaufmann, E. Aksan, J. Song, F. Pece, R. Ziegler, O. Hilliges
In Proceedings2020 International Conference on 3D Vision (3DV), 2020
ProjectCodePDFVideoBibTeX

Towards End-to-end Video-based Eye-Tracking

AuthorsS. Park, E. Aksan, X. Zhang, O. Hilliges
In ProceedingsEuropean Conference on Computer Vision (ECCV), 2020
ProjectCodePDFBibTeX
Teaser-Picture

Structured Prediction Helps 3D Human Motion Modelling

AuthorsE. Aksan*, M. Kaufmann*, O. Hilliges
In ProceedingsThe IEEE International Conference on Computer Vision (ICCV), Oct 2019
* These two authors contributed equally to this work.
ProjectCodePDFVideoBibTeX
Teaser-Picture

STCN:Stochastic Temporal Convolutional Networks

AuthorsE. Aksan, O. Hilliges
In ProceedingsInternational Conference on Learning Representations (ICLR), 2019
ProjectCodePDFBibTeX
Teaser-Picture

Deep Inertial Poser: Learning to Reconstruct Human Pose from Sparse Inertial Measurements in Real Time

AuthorsY. Huang*, M. Kaufmann*, E. Aksan, M. Black, O. Hilliges, G. Pons-Moll
In ACM Transactions on Graphics, (Proc. SIGGRAPH Asia), Nov 2018
* These two authors contributed equally to this work.
ProjectCodePDFVideoBibTeX
Teaser-Picture

DeepWriting: Making Digital Ink Editable via Deep Generative Modeling

AuthorsE. Aksan, F. Pece, O. Hilliges
In ProceedingsSIGCHI Conference on Human Factors in Computing Systems, Montréal, Canada, 2018

Honorable Mention Best Paper Award
ProjectCodePDFVideoBibTeX
Teaser-Picture

Learning human motion models for long-term predictions

AuthorsP. Ghosh, J. Song, E. Aksan, O. Hilliges
In Proceedings2017 International Conference on 3D Vision (3DV), 2017

Best Paper Award
ProjectCodePDFVideoBibTeX
Teaser-Picture

Guiding InfoGAN with Semi-Supervision

AuthorsA. Spurr, E. Aksan, O. Hilliges
In ProceedingsECML PKDD, Skopje, Macedonia, 2017
ProjectCodePDFBibTeX

Theses

An fMRI Segmentation Method Under Markov Random Fields for Brain Decoding — MSc Thesis, METU


Academic Activities

Reviewing

2022
CVPR, SIGGRAPH, TMLR, T-PAMI
2021
CVPR, ICCV, ICLR, NeurIPS (Outstanding Reviewer Award)
2020
CVPR, ECCV, ICLR, T-PAMI
2019
ICCV

Student Thesis

2020
MA Doruk Çetin Learning in-the-wild Temporal 3D Pose Estimation from MoCap Data
2019
MA Şahan Ayvaz A Study of Sparse Policy Networks for Deep Reinforcement Learning
MA Lukas Jendele Learning Functionally Decomposed Hierarchies for Continuous Navigation Tasks
MA Sami Hamdan Stochastic Temporal Convolutional Networks for Speech Enhancement
2018
MA Andreas Blöchliger Representation Learning for Sketch Suggestions based on the Combination of CNNs and RNNs
BA Şahan Ayvaz Emergence and Imitation of Locomotion in 2D and 3D Environments
BA Martin Blapp Evaluation of Human Motion Models
2017
MA Manuel Kaufmann A Deep Learning Approach to Human Motion Sequences Infilling
MA Adrian Spurr Semi-supervised Information Maximising Generative Adversarial Networks

Teaching