Yi Jiang
Yi Jiang has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Bingyang Wang
Dalian University of Technology, Dalian, China
1 shared publication - Chuang Lin
Department of Data Science, Monash University, Clayton, Australia
1 shared publication - Chuofan Ma
The University of Hong Kong, Hong Kong, China
1 shared publication Curtis Hsieh
GLOBALFOUNDRIES, Singapore
1 shared publication- Gholamreza Haffari
Monash University
1 shared publication - Hengshuang Zhao
The University of Hong Kong
1 shared publication
Latest Publications
- Top Cited Papers
- 7 January 2026
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 134 (1) , 39
Liquid: Language Models are Scalable and Unified Multi-Modal Generators
- 7 January 2026
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 134 (1) , 39
Top Cited PapersLiquid: Language Models are Scalable and Unified Multi-Modal Generators
- Open Access
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 2765-2774
A Simple Baseline for Open-World Tracking via Self-training
Open-World Tracking (OWT) presents a challenging yet emerging problem, aiming to track every object of any category. Different from traditional Multi-Object Tracking (MOT), OWT needs to additionally track targets beyond predefined categories in the training set. To address the problem, we propose a simple baseline, SimOWT. We simplify the recently proposed OWT algorithm by streamlining the association module and accelerating the inference speed. By leveraging the self-training paradigm, SimOWT can distinguish unknown-class targets from the background, fully unleashing the potential of TAO-OW dataset. Furthermore, we enhance SimOWT from the perspectives of Pseudo Boxes Merging and Re-Weighting, thereby discovering more targets belonging to unknown classes and reducing the sensitivity of the model to low-quality pseudo-labels. Benefiting from the proposed approaches, SimOWT demonstrates a significant improvement in tracking performance on unknown classes. Moreover, the comprehensive experiments on the TAO-OW benchmark demonstrate that our model outperforms the state-of-the-art OWT method, OWTB, with an absolute gain of 11.2% OWTA and 16.4% detection recall respectively on unknown classes. The code is released at https://github.com/22109095/SimOWT.
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 2765-2774
Open AccessA Simple Baseline for Open-World Tracking via Self-training
Open-World Tracking (OWT) presents a challenging yet emerging problem, aiming to track every object of any category. Different from traditional Multi-Object Tracking (MOT), OWT needs to additionally track targets beyond predefined categories in the training set. To address the problem, we propose a simple baseline, SimOWT. We simplify the recently proposed OWT algorithm by streamlining the association module and accelerating the inference speed. By leveraging the self-training paradigm, SimOWT can distinguish unknown-class targets from the background, fully unleashing the potential of TAO-OW dataset. Furthermore, we enhance SimOWT from the perspectives of Pseudo Boxes Merging and Re-Weighting, thereby discovering more targets belonging to unknown classes and reducing the sensitivity of the model to low-quality pseudo-labels. Benefiting from the proposed approaches, SimOWT demonstrates a significant improvement in tracking performance on unknown classes. Moreover, the comprehensive experiments on the TAO-OW benchmark demonstrate that our model outperforms the state-of-the-art OWT method, OWTB, with an absolute gain of 11.2% OWTA and 16.4% detection recall respectively on unknown classes. The code is released at https://github.com/22109095/SimOWT.
- 29 October 2022
- Published by Springer Nature
- p. 380-397
Multimodal Transformer with Variable-Length Memory for Vision-and-Language Navigation
- 29 October 2022
- Published by Springer Nature
- p. 380-397
Multimodal Transformer with Variable-Length Memory for Vision-and-Language Navigation
- 5 October 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 61-63
Analysis of Extreme Pattern Density and Interconnects Routing Impacts on Via Open
Two phenomena of dual damascene via open in extremely low local metal pattern density region, and when connected through a narrow width metal line to a big pad opening with high density sea of vias are identified. A systematic study was carried out to investigate the impact of extreme pattern density design on dual damascene via integrity. Local metal density, interconnects metal line width, length, and pad size connecting to an isolated via are investigated to establish the correlation to via open and to explain the failure mechanism. Finally design for manufacturing solution is proposed, apart from process fix as the total solution.
- 5 October 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 61-63
Analysis of Extreme Pattern Density and Interconnects Routing Impacts on Via Open
Two phenomena of dual damascene via open in extremely low local metal pattern density region, and when connected through a narrow width metal line to a big pad opening with high density sea of vias are identified. A systematic study was carried out to investigate the impact of extreme pattern density design on dual damascene via integrity. Local metal density, interconnects metal line width, length, and pad size connecting to an isolated via are investigated to establish the correlation to via open and to explain the failure mechanism. Finally design for manufacturing solution is proposed, apart from process fix as the total solution.