Bo Dong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Bo Dong has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Chuanfei Hu
Faculty of Intelligence Technology, Shanghai Institute of Technology, Shanghai, 201418, China
1 shared publication - Fan Wang
Alibaba Group
1 shared publication - Hang Shao
School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China
1 shared publication - Hao Luo
DAMO Academy, Alibaba Group, Hangzhou, China
1 shared publication - Huan Xiong
Institute for Advanced Study in Mathematics Harbin Institute of Technology 150001 Heilongjiang P.R. China
1 shared publication - Jialun Pei
Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong, SAR, China
1 shared publication
Latest Publications
- Open Access
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 6598-6606
Adaptive Query Selection for Camouflaged Instance Segmentation
Camouflaged instance segmentation is a challenging task due to the various aspects such as color, structure, lighting, etc., of object instances embedded in complex backgrounds. Although the current DETR-based scheme simplifies the pipeline, it suffers from a large number of object queries, leading to many false positive instances. To address this issue, we propose an adaptive query selection mechanism. Our research reveals that a large number of redundant queries scatter the extracted features of the camouflaged instances. To remove these redundant queries with weak correlation, we evaluate the importance of the object query from the perspectives of information entropy and volatility. Moreover, we observed that occlusion and overlapping instances significantly impact the accuracy of the selection mechanism. Therefore, we design a boundary location embedding mechanism that incorporates fake instance boundaries to obtain better location information for more accurate query instance matching. We conducted extensive experiments on two challenging camouflaged instance segmentation datasets, namely COD10K and NC4K, and demonstrated the effectiveness of our proposed model. Compared with the OSFormer, our model significantly improves the performance by 3.8% AP and 5.6% AP with less computational cost, achieving the state-of-the-art of 44.8 AP and 48.1 AP with ResNet-50 on the COD10K and NC4K test-dev sets, respectively.
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 6598-6606
Open AccessAdaptive Query Selection for Camouflaged Instance Segmentation
Camouflaged instance segmentation is a challenging task due to the various aspects such as color, structure, lighting, etc., of object instances embedded in complex backgrounds. Although the current DETR-based scheme simplifies the pipeline, it suffers from a large number of object queries, leading to many false positive instances. To address this issue, we propose an adaptive query selection mechanism. Our research reveals that a large number of redundant queries scatter the extracted features of the camouflaged instances. To remove these redundant queries with weak correlation, we evaluate the importance of the object query from the perspectives of information entropy and volatility. Moreover, we observed that occlusion and overlapping instances significantly impact the accuracy of the selection mechanism. Therefore, we design a boundary location embedding mechanism that incorporates fake instance boundaries to obtain better location information for more accurate query instance matching. We conducted extensive experiments on two challenging camouflaged instance segmentation datasets, namely COD10K and NC4K, and demonstrated the effectiveness of our proposed model. Compared with the OSFormer, our model significantly improves the performance by 3.8% AP and 5.6% AP with less computational cost, achieving the state-of-the-art of 44.8 AP and 48.1 AP with ResNet-50 on the COD10K and NC4K test-dev sets, respectively.
- Open Access
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 2131-2138
A Unified Query-based Paradigm for Camouflaged Instance Segmentation
Due to the high similarity between camouflaged instances and the background, the recently proposed camouflaged instance segmentation (CIS) faces challenges in accurate localization and instance segmentation. To this end, inspired by query-based transformers, we propose a unified query-based multi-task learning framework for camouflaged instance segmentation, termed UQFormer, which builds a set of mask queries and a set of boundary queries to learn a shared composed query representation and efficiently integrates global camouflaged object region and boundary cues, for simultaneous instance segmentation and instance boundary detection in camouflaged scenarios. Specifically, we design a composed query learning paradigm that learns a shared representation to capture object region and boundary features by the cross-attention interaction of mask queries and boundary queries in the designed multi-scale unified learning transformer decoder. Then, we present a transformer-based multi-task learning framework for simultaneous camouflaged instance segmentation and camouflaged instance boundary detection based on the learned composed query representation, which also forces the model to learn a strong instance-level query representation. Notably, our model views the instance segmentation as a query-based direct set prediction problem, without other post-processing such as non-maximal suppression. Compared with 14 state-of-the-art approaches, our UQFormer significantly improves the performance of camouflaged instance segmentation. Our code will be available at: https://github.com/dongbo811/UQFormer.
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 2131-2138
Open AccessA Unified Query-based Paradigm for Camouflaged Instance Segmentation
Due to the high similarity between camouflaged instances and the background, the recently proposed camouflaged instance segmentation (CIS) faces challenges in accurate localization and instance segmentation. To this end, inspired by query-based transformers, we propose a unified query-based multi-task learning framework for camouflaged instance segmentation, termed UQFormer, which builds a set of mask queries and a set of boundary queries to learn a shared composed query representation and efficiently integrates global camouflaged object region and boundary cues, for simultaneous instance segmentation and instance boundary detection in camouflaged scenarios. Specifically, we design a composed query learning paradigm that learns a shared representation to capture object region and boundary features by the cross-attention interaction of mask queries and boundary queries in the designed multi-scale unified learning transformer decoder. Then, we present a transformer-based multi-task learning framework for simultaneous camouflaged instance segmentation and camouflaged instance boundary detection based on the learned composed query representation, which also forces the model to learn a strong instance-level query representation. Notably, our model views the instance segmentation as a query-based direct set prediction problem, without other post-processing such as non-maximal suppression. Compared with 14 state-of-the-art approaches, our UQFormer significantly improves the performance of camouflaged instance segmentation. Our code will be available at: https://github.com/dongbo811/UQFormer.
- 28 February 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 72 (00189456) , 1-11
Toward Purifying Defect Feature for Multilabel Sewer Defect Classification
An automatic vision-based sewer inspection plays a key role of sewage system in a modern city. Recent advances focus on utilizing a deep learning model to realize the sewer inspection system, benefiting from the capability of data-driven feature extraction. However, the ambiguity of sewer defects in the feature space is ignored, deteriorating the performance of sewer inspection. There are two reasons for such ambiguity. First, the defect-irrelevant region interferes the feature extraction of the model. Second, the setting of multilabel is an inherent challenge of extracting discriminative feature for different defect classes from defect-relevant region. In this article, we propose a multilabel sewer defect classification method, which can purify the sewer defect components in latent spaces, thereby mitigating the ambiguity of sewer defects. Specifically, a novel self-purification module (SPM) is modeled to disentangle the ambiguity of sewer defect feature, which consists of intraclass purification (ICP) and interclass decorrelation (ICD). ICP utilizes the task-aware information to purify the sewer features, and ICD aims to eliminate the cross correlation and defect-irrelevant components simultaneously. Moreover, to ensure the reliability of SPM, center global alignment (CGA) is introduced to avoid the trivial solution. Experimental results demonstrate the superiority of the proposed method compared with seven state-of-the-art methods on the latest benchmark Sewer-ML. The proposed method outperforms the others approximately 8% $\mathrm {F2_{\text {CIW}}}$ with a tolerant inference speed. Finally, the robustness of the proposed method is verified in two perturbed scenarios, where the reliable performance can be guaranteed against the limited perturbations.
- 28 February 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 72 (00189456) , 1-11
Toward Purifying Defect Feature for Multilabel Sewer Defect Classification
An automatic vision-based sewer inspection plays a key role of sewage system in a modern city. Recent advances focus on utilizing a deep learning model to realize the sewer inspection system, benefiting from the capability of data-driven feature extraction. However, the ambiguity of sewer defects in the feature space is ignored, deteriorating the performance of sewer inspection. There are two reasons for such ambiguity. First, the defect-irrelevant region interferes the feature extraction of the model. Second, the setting of multilabel is an inherent challenge of extracting discriminative feature for different defect classes from defect-relevant region. In this article, we propose a multilabel sewer defect classification method, which can purify the sewer defect components in latent spaces, thereby mitigating the ambiguity of sewer defects. Specifically, a novel self-purification module (SPM) is modeled to disentangle the ambiguity of sewer defect feature, which consists of intraclass purification (ICP) and interclass decorrelation (ICD). ICP utilizes the task-aware information to purify the sewer features, and ICD aims to eliminate the cross correlation and defect-irrelevant components simultaneously. Moreover, to ensure the reliability of SPM, center global alignment (CGA) is introduced to avoid the trivial solution. Experimental results demonstrate the superiority of the proposed method compared with seven state-of-the-art methods on the latest benchmark Sewer-ML. The proposed method outperforms the others approximately 8% $\mathrm {F2_{\text {CIW}}}$ with a tolerant inference speed. Finally, the robustness of the proposed method is verified in two perturbed scenarios, where the reliable performance can be guaranteed against the limited perturbations.