Research
His research focuses on learning discriminative, robust and generalizable visual representations for multiple scenarios such as autonomous driving, medical imaging and aerial imaging. Some representative publications are listed below.
|
|
Learning Content-enhanced Mask Transformer for Domain Generalized Urban-scene Segmentation
Qi Bi, Shaodi You, Theo Gevers
Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2024
code
Learning domain generalized scene segmentation by content-enhanced mask attention mechanism.
|
|
Learning Generalized Segmentation for Foggy-Scenes by Bi-directional Wavelet Guidance
Qi Bi, Shaodi You, Theo Gevers
Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2024
code
Learning scene segmentation that can be generalized to arbitrary unseen foggy target domains from only a clear source domain; the first work for this task.
|
|
Interactive Learning of Intrinsic and Extrinsic Properties for All-day Semantic Segmentation
Qi Bi, Shaodi You, Theo Gevers
IEEE Transactions on Image Processing (T-IP), 2023
code
dataset
Learning robust scene semantic segmentation under all-day scenarios; proposing the first all-day semantic segmentation dataset All-day CityScapes.
|
|
Learning Generalized Medical Image Segmentation from Decoupled Feature Queries
Qi Bi, Jingjun Yi, Hao Zheng, Wei Ji, Yawen Huang, Yuexiang Li, Yefeng Zheng
Proceedings of the AAAI Conference on Artificial Intelligence (AAAI)
code
Learning domain generalized medical image segmentation by querying from decoupled features; the first work to leverage Vision Transformer for domain generalized medical image segmentation.
|
|
Segment anything is not always perfect: An investigation of sam on different real-world applications
Wei Ji, Jingjing Li, Qi Bi, Wenbo Li, Li Cheng
CVPR 1st workshop on Vision-based InduStrial InspectiON, 2023
Best paper award
code
Benchmarking Segment Anything (SAM) on multiple real-world scenarios.
|
|
All Grains, One Scheme (AGOS): Learning Multi-grain Instance Representation for Aerial Scene Classification
Qi Bi, Beichen Zhou, Kun Qin, Qinghao Ye, Gui-Song Xia
IEEE Transactions on Geoscience and Remote Sensing (T-GRS), 2022
ESI highly-cited paper
code
Extending deep multiple instance learning into a multi-grain framework while maintaining the same semantic scheme, dubbed as AGOS; learning discriminative aerial scene representation by AGOS.
|
|
Label-efficient Hybrid-supervised Learning for Medical Image Segmentation
Junwen Pan*, Qi Bi*, Yanzhan Yang, Pengfei Zhu, Cheng Bian
* : equal contribution
Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2022
Learning weakly semi-supervised medical image segmentation by the proposed dynamic instance indicator and dynamic co-regularization framework.
|
|
Local semantic enhanced convnet for aerial scene recognition
Qi Bi, Kun Qin, Han Zhang, Gui-Song Xia
IEEE Transactions on Image Processing (T-IP), 2021
ESI highly-cited paper
code
Learning aerial scene representation by modeling context-aware class peak response.
|
|
Joint semantic mining for weakly supervised RGB-D salient object detection
Jingjing Li, Wei Ji, Qi Bi, Cheng Yan, Miao Zhang, Yongri Piao, Huchuan Lu
Advances in Neural Information Processing Systems (NeurIPS), 2021
code
dataset
Learning weakly-supervised RGB-D salient object detection (SOD) from the image, depth map and image caption; proposing a dataset for caption based SOD dubbed as CapS.
|
|
Local-global dual perception based deep multiple instance learning for retinal disease classification
Qi Bi, Shuang Yu, Wei Ji, Cheng Bian, Lijun Gong, Hanruo Liu, Kai Ma, Yefeng Zheng
Medical Image Computing and Computer Assisted Intervention (MICCAI), 2021
MICCAI2021 travel awards
MICCAI2021 young scientist award candidate
Learning retinal diseases from fundus images by local-global representation.
|
|
MIL-ViT: A multiple instance vision transformer for fundus image classification
Qi Bi, Xu Sun, Shuang Yu, Kai Ma, Cheng Bian, Munan Ning, Nanjun He, Yawen Huang, Yuexiang Li, Hanruo Liu, Yefeng Zheng
Medical Image Computing and Computer Assisted Intervention (MICCAI), 2021, conference version
Journal of Visual Communication and Image Representation, 2023, journal version
code
Learning medical image Transformer by deep multiple instance learning.
|
|
Learning calibrated medical image segmentation via multi-rater agreement modeling
Wei Ji, Shuang Yu, Junde Wu, Kai Ma, Cheng Bian, Qi Bi, Jingjing Li, Hanruo Liu, Li Cheng, Yefeng Zheng
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021
Best paper candidate
code
Learning medical image segmentation from multiple annotations by multi-rater modeling.
|
|
A multiple-instance densely-connected ConvNet for aerial scene classification
Qi Bi, Kun Qin, Zhili Li, Han Zhang, Kai Xu, Gui-Song Xia
IEEE Transactions on Image Processing (T-IP), 2020
ESI highly-cited paper
code
Modeling discriminative aerial scene representation by deep multiple instance learning.
|
Supervision
Noud Corten, Improved Road Crack Severity Measurement Using Deep Convolutional Networks by Storing Spatial Information, November 2021-August 2022 (completed).
Carlo Airaghi, Multi-Stage Multiscale Training Architecture for Semantic Segmentation of Remote Sensing Images, April 2021- December 2021 (completed).
Silvan Murre, Layout2Land: Semi-Supervised Learning of a Layout and Style Reconfigurable GAN, March 2021-June 2021 (completed).
|
Teaching
2024 Vision & Autonomous Robotics (UvA, Lecturer)
2024 Computer Vision 1 (UvA, Lecturer)
2024 Computer Vision 2 (UvA, Teaching Assistant)
2023 Computer Vision 1 (UvA, Teaching Assistant)
2023 Computer Vision 2 (UvA, Teaching Assistant)
2022 Computer Vision 1 (UvA, Teaching Assistant)
2021 Computer Vision 1 (UvA, Teaching Assistant)
2020 Computer Vision 1 (UvA, Teaching Assistant)
|
|