E-Mail: ddai [at] mpi-inf.mpg.de
Phone: +49 681 9325 2104
Address: MPI for Informatics, Saarland Informatics Campus, 66123 Saarbrücken
Room: E1 4 – 604
a novel test-time domain adaptation method for depth estimation
a very simple yet effective method for object detection across domains
IJCV (under review), 2022
a comprehensive study on the robustness of LiDAR semantic segmentation methods
a unified and learning based approach to the 3D MOT problem
a novel pipeline to generate Neural Radiance Fields (NeRF) of an object or a scene of a specific class, conditioned on a single input image
a novel co-learning framework (CoSSL) with decoupled representation learning and classifier learning for imbalanced SSL
The first end-to-end approach to learn to optimize the LiDAR beam configuration for given applications
The first MOT formulation designed to be solved with Adiabatic Quantum Computing.
CVPR (Oral), 2022
A novel simulation approach to simulate snowfall effects into existing LiDAR dataset to train robust LiDAR-based perception methods for adverse weather
A novel method for long-term test-time adaptation under continually changing environments.
ZegFormer is the first framework that decouple the zero-shot semantic segmentation into: 1) class-agnostic segmentation and 2) segment-level zero-shot classification
CVPR (Oral), 2022
ScribbleKITTI, the first scribble-annotated dataset for LiDAR semantic segmentation and a novel learning method to reduce the performance gap that arises when using such weak annotations.
A multi-resolution training approach for UDA, that combines the strengths of small high-resolution crops to preserve fine segmentation details and large low-resolution crops to capture long-range context dependencies with a learned scale attention.
A novel Motion TRansformer (MTR) framework that models motion prediction as the joint optimization of global intention localization and local movement refinement. Won the championship of Waymo Open Dataset Challenge 2022 on Motion Prediction
We present Scale-aware Domain Adaptive Faster R-CNN, a model aiming at improving the cross-domain robustness of object detection.
We propose a projection-based method for semantic segmentation of LiDar data, called Multi-scale Interaction Network (MINet), which is very efficient and accurate.
This work develops an approach for scene understanding purely based on binaural sounds.
We present a domain flow generation (DLOW) model to bridge two different domains by generating a continuous sequence of intermediate domains flowing from one domain to the other.
We introduce Task Switching Networks (TSNs), a task-conditioned architecture with a single unified encoder/decoder for efficient multi-task learning.
DAFormer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation
A novel UDA method, DAFormer, consisting of a Transformer encoder and a multi-level context-aware feature fusion decoder, improve SOTA by 10.8 mIoU for GTA->Cityscapes and 5.4 mIoU for Synthia->Cityscapes
We propose a novel co-learning framework (CoSSL) with decoupled representation learning and classifier learning for imbalanced SSL.
Zero-shot semantic segmentation (ZS3) aims to segment the novel categories that have not been seen in the training. We propose to decouple the ZS3 into two sub-tasks: 1) a class-agnostic grouping task to group the pixels into segments. 2) a zero-shot classification task on segments.
This work presents a novel method for LiDAR-based 3D object detection in foggy weather by simulating foggy effects into standard LiDAR data..
ACDC is a large-scale dataset for training and testing semantic segmentation methods for four adverse visual conditions: fog, nighttime, rain, and snow.
A novel method to leverage the guidance from self-supervised depth estimation to bridge the domain gap for semantic segmentation, achieving state-of-the-art performance..
A novel manner to learn end-to-end driving from a reinforcement learning expert that maps bird's-eye view images to continuous low-level actions, achieving the state-of-the-art perfomrance.
A novel method to combine and reuse existing datasets that belong to different domains, have partial annotations, and/or have different data modalities, to boost the overall performance on the target domain.
We propose a novel end-to-end driving method that can learn how-to-drive directly from surrounding-view videos and route planners.
- Associate Editor of IJCV
- Area Chair of WACV 2020, CVPR 2021, CVPR 2022, ECCV 2020, and ICRA 2022.
- Lead workshop organizer: Vision for All Seasons at CVPR’19, CVPR’20, CVPR’21, CVPR’22.
- Co-organizer of Workshop DeepMTL: Deep Multi-Task Learning at ICCV’21.
- Lead Guest Editor of the Special Issue “Computer Vision for All Seasons” of IJCV.
- Senior Program Committee Member: AAAI 2020 and IJCAI 2019.
- Lead workshop organizer: Autonomous Driving workshop at ICCV’19.
- Main organizer of challenge Learning to Drive at ICCV’19
- Main organizer of challenge Nighttime Semantic Image Segmentation at CVPR’20.
- Jury member for the Pioneer Fellowship Program at ETH Zurich.
- Examiner of Doctoral Exams: EPFL, NUS, ETH Zurich, and Inria.
- Regular Reviewers: CVPR, ICCV, ECCV, NeurIPS, ICLR, ICML, ICRA, IROS, AAAI, IJCV, and PAMI.
- The ACM Computer Science in Cars Symposium, 2022
- The“3D Perception for Autonomous Driving” workshop at ECCV,2022
- TU Vienna, August 2022
- The German Conference on PatternRecognition, 2022
- The IV workshop “Beyond supervised learning: addressing data scarcity in intelligent transportation systems“, 2022
- The “Learning to Understand Aerial Images (LUAI)” Workshop at ICCV, Oct. 2021.
- The Saarland Informatics Campus Lecture Series, Oct. 2021.
- The “Radar Perception for All-Weather Autonomy” Workshop at ICRA, May 2021.
- Zurich High School, “General Introduction to Autonomous Driving”, Nov 2020.
- The Robotics and Perception Group at the University of Zurich, Sep. 2020.
- The “Transferring and Adapting Source Knowledge in ComputerVision and VISDA Challenge” Workshop at ECCV, 2020.
- The “Map-based Localization for Autonomous Driving” Workshop at ECCV, 2020.
- The “Commands for Autonomous Vehicle” Workshop at ECCV, 2020.
- The “Autonomous Driving Workshop” at CVPR, 2020.
- The “Bridging the Gap between Computational Photography and Visual Recognition” Workshop at CVPR, 2020.
- The “Physics-Based Vision meets Deep Learning (PBDL)” Workshop at ICCV, 2019.
- International VDI Conference – Future of AI in Automotive, Berlin, 2018.
- The “ApolloScape: the Vision-based Navigation for Autonomous Driving” Workshop at ECCV, 2018.
- Scientifica: Zurich Science Days, 2017.
Deep Learning for Autonomous Driving, 6 ECTS, ETH Zurich, 2020 (80+ students).
- Deep Learning for Autonomous Driving, 6 ECTS, ETH Zurich, 2021 (100+ students).
- Deep Learning for Autonomous Driving, 6 ECTS, ETH Zurich, 2022 (100+ students).
- Tutorial “From Image Restoration to Enhancement and Beyond” at ICCV, 2019.