| Website | collaborative-ai.org |
Excited to share that our paper โRobustSpring: Benchmarking Robustness to Image Corruptions for Optical Flow, Scene Flow and Stereoโ has been accepted to #ICLR2026 ๐.
We introduce RobustSpring, a new benchmark that evaluates not only accuracy but also robustness of optical flow, scene flow, and stereo models under 20 realโworld image corruptions.
Congratulations to the authors!
๐ For more news: https://www.collaborative-ai.org/publications/oei26_iclr/
Which part of graphs do people look at when solving analytical tasks?
๐ฐ Our work "Towards a Better Understanding of Graph Perception in Immersive Environments" was accepted to Graph Drawing #GD2025.
Congratulations to the authors!
Learn more about this work from our website: https://www.collaborative-ai.org/publications/zhang25_gd/
๐ Exciting News! ๐
HOIGaze: Gaze Estimation During Hand-Object Interactions in Extended Reality has been accepted to #SIGGRAPH 2025! ๐
HOIGaze introduces:1๏ธโฃ A hierarchical framework that first identifies which hand the user is visually attending to, then estimates gaze direction based on the handโs posture. 2๏ธโฃ A gaze estimation network that combines graph neural networks and cross-modal Transformers. 3๏ธโฃ An eye-head coordination loss function.
๐ Learn more: https://collaborative-ai.org/publications/hu25_siggraph/
๐ Exciting News! ๐
We're thrilled to share that our group's proposal, "๐ ๐๐น๐๐ถ๐ ๐ฒ๐ฑ๐ถ๐ฎ๐๐ฒ: ๐ ๐๐น๐๐ถ๐บ๐ผ๐ฑ๐ฎ๐น ๐๐ฒ๐ต๐ฎ๐๐ถ๐ผ๐๐ฟ ๐๐ป๐ฎ๐น๐๐๐ถ๐ ๐ณ๐ผ๐ฟ ๐๐ฟ๐๐ถ๐ณ๐ถ๐ฐ๐ถ๐ฎ๐น ๐ ๐ฒ๐ฑ๐ถ๐ฎ๐๐ถ๐ผ๐ป", has been accepted as a Grand Challenge at ACM Multimedia 2025!๐
"The goal of this multi-year challenge is to contribute to realising the vision of autonomous artificial mediators by measurable advances in key conversational behaviour sensing and analysis tasks."
For more details, visit https://www.multimediate-challenge.org/.
๐ Exciting News! ๐ Our groupโs paper ๐ฉยฒ๐๐ถ๐ฎ๐น: ๐จ๐ป๐ถ๐ณ๐ถ๐ฐ๐ฎ๐๐ถ๐ผ๐ป ๐ผ๐ณ ๐ฉ๐ถ๐ฑ๐ฒ๐ผ ๐ฎ๐ป๐ฑ ๐ฉ๐ถ๐๐๐ฎ๐น ๐๐ถ๐ฎ๐น๐ผ๐ด ๐๐ถ๐ฎ ๐ ๐๐น๐๐ถ๐บ๐ผ๐ฑ๐ฎ๐น ๐๐ ๐ฝ๐ฒ๐ฟ๐๐ has been accepted at CVPR 2025! ๐๐
VยฒDial is a novel model specifically designed to handle both image and video input data for multimodal conversational tasks. Extensive evaluations on AVSD and VisDial datasets show that VยฒDial achieves new state-of-the-art results across multiple benchmarks.
Congratulations to the authors. ๐
๐ข New Paper Alert! ๐ข
We're thrilled to announce that our paper, HAIFAI: Human-AI Interaction for Mental Face Reconstruction, has been accepted by ACM Transactions on Interactive Intelligent Systems (TiiS).
Congratulations to the authors!
You can check out the preprint on arXiv https://arxiv.org/abs/2412.06323v1 and stay tuned for the camera-ready version on our website https://collaborative-ai.org/.
We present HAIFAI - a novel collaborative human-AI system to tackle the challenging task of reconstructing a visual representation of a face that exists only in a person's mind. Users iteratively rank images presented by the AI system based on their resemblance to a mental image. These rankings, in turn, allow the system to extract relevant image features, fuse them into a unified feature vector, and use a generative model to reconstruct the mental image. We also propose an extension called HAIFAI-X that allows users to manually refine and further improve the reconstruction using an easy-to-use slider interface. To avoid the need for tedious human data collection for model training, we introduce a computational user model of human ranking behaviour. For this, we collected a small face ranking dataset through an online crowd-sourcing study containing data from 275 participants. We evaluate HAIFAI and HAIFAI-X in a 12-participant user study and show that HAIFAI outperforms the previous state of the art regarding reconstruction quality, usability, perceived workload, and reconstruction speed. HAIFAI-X achieves even better reconstruction quality at the cost of reduced usability, perceived workload, and increased reconstruction time. We further validate the reconstructions in a subsequent face ranking study with 18 participants and show that HAIFAI-X achieves a new state-of-the-art identification rate of 60.6%. These findings represent a significant advancement towards developing new collaborative intelligent systems capable of reliably and effortlessly reconstructing a user's mental image.
๐ Exciting News! ๐
Our group has two papers (conditionally) accepted to #CHI2025.
1๏ธโฃ SummAct: Uncovering User Intentions Through Interactive Behaviour Summarisation
2๏ธโฃ How People Read Charts: A Model of Task-driven Eye Movement Control
Stay tuned for more details about the papers at: https://collaborative-ai.org/news/2025/01/two-paper-accepted-at-chi/
๐We are thrilled to share that our lab director Prof. @abulling has joined the editorial board of the IEEE Transactions on Visualization and Computer Graphics (TVCG) as of January 1, 2025.
Congratulations, Prof. Bulling! ๐
๐ Exciting News! ๐
Our lab is a part of the ICRA 2025 workshop on Nonverbal Cues for Human-Robot Cooperative Intelligence! ๐คโจ
This workshop is organized by an incredible team, including our own alumnus Prof. Xucong Zhang and lab director Prof. Andreas Bulling @abulling.
Stay tuned for more updates from this exciting event! ๐
We are excited to announce that the funding proposal "Privacy-preserving Eye Tracking" has been approved by the German Research Foundation (DFG)! ๐
It is a joint proposal with Prof. Ralf Kรผsters at the Institute of Information Security, and will fund one PostDoc and one PhD position for three years.
The goal of this project is to advance privacy-preserving eye tracking along the whole eye tracking data processing pipeline.
