Xiaobin Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xiaobin Li has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Hongxu Jiang
Beihang University
4 shared publications - Fangzheng Tian
School of Computer Science and Engineering, Beihang University, Beijing 100191, China
3 shared publications - Runhua Zhang
School of Artificial Intelligence, Beihang University, Beijing, China
2 shared publications Baozhi Wei
School of Beihang University, Beijing, China
1 shared publication- Chenhui Zhu
School of Computer Science and Engineering, Beihang University, Beijing 100191, China
1 shared publication - Chenxu Wang1 shared publication
Latest Publications
- 7 February 2026
- Published by Elsevier in Signal Processing
- Vol. 244
DRA-Net: Dynamic region-adaptive kernel generation for spectrally faithful pansharpening
- 7 February 2026
- Published by Elsevier in Signal Processing
- Vol. 244
DRA-Net: Dynamic region-adaptive kernel generation for spectrally faithful pansharpening
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35, 7319-7331
ELBO-T2IAlign: A Generic ELBO-Based Method for Calibrating Pixel-Level Text-Image Alignment in Diffusion Models
Diffusion models excel at image generation. Recent studies have shown that these models not only generate high-quality images but also encode text-image alignment information through attention maps or loss functions. This information is valuable for various downstream tasks, including segmentation, text-guided image editing, and compositional image generation. However, current methods heavily rely on the assumption of perfect text-image alignment in diffusion models, which is not the case. In this paper, we propose using zero-shot referring image segmentation as a proxy task to evaluate the pixel-level image and class-level text alignment of popular diffusion models. We conduct an in-depth analysis of pixel-text misalignment in diffusion models from the perspective of training data bias. We find that misalignment occurs in images with small-sized, occluded, or rare object classes. Therefore, we propose ELBO-T2IAlign-a simple yet effective method to calibrate pixel-text alignment in diffusion models based on the evidence lower bound (ELBO) of likelihood. ELBO-T2IAlign is training-free and generic: it requires no additional annotations, model retraining, or architectural modifications, and it can be directly applied to different diffusion backbones. Extensive experiments on zero-shot referring image segmentation, text-guided image editing, and compositional image generation verify that the proposed calibration improves pixel-text alignment across complementary downstream tasks.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35, 7319-7331
ELBO-T2IAlign: A Generic ELBO-Based Method for Calibrating Pixel-Level Text-Image Alignment in Diffusion Models
Diffusion models excel at image generation. Recent studies have shown that these models not only generate high-quality images but also encode text-image alignment information through attention maps or loss functions. This information is valuable for various downstream tasks, including segmentation, text-guided image editing, and compositional image generation. However, current methods heavily rely on the assumption of perfect text-image alignment in diffusion models, which is not the case. In this paper, we propose using zero-shot referring image segmentation as a proxy task to evaluate the pixel-level image and class-level text alignment of popular diffusion models. We conduct an in-depth analysis of pixel-text misalignment in diffusion models from the perspective of training data bias. We find that misalignment occurs in images with small-sized, occluded, or rare object classes. Therefore, we propose ELBO-T2IAlign-a simple yet effective method to calibrate pixel-text alignment in diffusion models based on the evidence lower bound (ELBO) of likelihood. ELBO-T2IAlign is training-free and generic: it requires no additional annotations, model retraining, or architectural modifications, and it can be directly applied to different diffusion backbones. Extensive experiments on zero-shot referring image segmentation, text-guided image editing, and compositional image generation verify that the proposed calibration improves pixel-text alignment across complementary downstream tasks.
- 24 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Systems Journal
- Vol. 19 (2) , 553-564
A High-Dimensional Dynamic Characteristics-Based Boundary Test Scenario Generation Method for UAVs in Dynamic Target Tracking
Effective boundary scenario generation methods are critical for dynamic target tracking when evaluating the performance of unmanned aerial vehicles. However, traditional methods primarily rely on exhaustive testing or random sampling and often assume that different contributing factors are independent. This results in the ineffective generation of boundary scenarios and a lack of realism. In this study, a novel high-dimensional dynamic characteristics-based boundary test scenario generation method is proposed using reinforcement learning (RL). Boundary test scenarios can be explored more purposive, by designing an appropriate reward function. First, an innovative scenario modeling method is developed to model the influence of environment, occlusion, and other interference. A joint distribution model of the key correlation factors, such as light intensity and clouds, is also established. This ensures scenario authenticity and test accuracy. Subsequently, a high-dimensional dynamic spatial Markov decision process (HDDS-MDP) model is constructed to facilitate the generation of boundary scenarios based on the scenario modeling. Ultimately, RL is employed to solve the HDDS-MDP model and generate boundary test scenario, thereby substantially improves the effectiveness of boundary test scenario generation. The simulation results indicate that the boundary test scenario generation effectiveness is improved by 91% over random sampling methods.
- 24 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Systems Journal
- Vol. 19 (2) , 553-564
A High-Dimensional Dynamic Characteristics-Based Boundary Test Scenario Generation Method for UAVs in Dynamic Target Tracking
Effective boundary scenario generation methods are critical for dynamic target tracking when evaluating the performance of unmanned aerial vehicles. However, traditional methods primarily rely on exhaustive testing or random sampling and often assume that different contributing factors are independent. This results in the ineffective generation of boundary scenarios and a lack of realism. In this study, a novel high-dimensional dynamic characteristics-based boundary test scenario generation method is proposed using reinforcement learning (RL). Boundary test scenarios can be explored more purposive, by designing an appropriate reward function. First, an innovative scenario modeling method is developed to model the influence of environment, occlusion, and other interference. A joint distribution model of the key correlation factors, such as light intensity and clouds, is also established. This ensures scenario authenticity and test accuracy. Subsequently, a high-dimensional dynamic spatial Markov decision process (HDDS-MDP) model is constructed to facilitate the generation of boundary scenarios based on the scenario modeling. Ultimately, RL is employed to solve the HDDS-MDP model and generate boundary test scenario, thereby substantially improves the effectiveness of boundary test scenario generation. The simulation results indicate that the boundary test scenario generation effectiveness is improved by 91% over random sampling methods.
- 6 April 2023
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 79 (13) , 14313-14341
LOCP: Latency-optimized channel pruning for CNN inference acceleration on GPUs
- 6 April 2023
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 79 (13) , 14313-14341
LOCP: Latency-optimized channel pruning for CNN inference acceleration on GPUs
- 25 March 2022
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 78 (11) , 13780-13813
High anti-interference and FPGA-oriented method for real-time ship detection based on structured LBP features
- 25 March 2022
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 78 (11) , 13780-13813
High anti-interference and FPGA-oriented method for real-time ship detection based on structured LBP features
- 1 September 2021
- Published by Elsevier in Neurocomputing
- Vol. 455, 12-22
Robustness-aware 2-bit quantization with real-time performance for neural network
- 1 September 2021
- Published by Elsevier in Neurocomputing
- Vol. 455, 12-22
Robustness-aware 2-bit quantization with real-time performance for neural network
- Open Access
Robustness-aware 2-bit quantization with real-time performance for neural network
Quantized neural network (NN) with a reduced bit precision is an effective solution to reduces the computational and memory resource requirements and plays a vital role in machine learning. However, it is still challenging to avoid the significant accuracy degradation due to its numerical approximation and lower redundancy. In this paper, a novel robustness-aware 2-bit quantization scheme is proposed for NN base on binary NN and generative adversarial network(GAN), witch improves the performance by enriching the information of binary NN, efficiently extract the structural information and considering the robustness of the quantized NN. Specifically, using shift addition operation to replace the multiply-accumulate in the quantization process witch can effectively speed the NN. Meanwhile, a structural loss between the original NN and quantized NN is proposed to such that the structural information of data is preserved after quantization. The structural information learned from NN not only plays an important role in improving the performance but also allows for further fine tuning of the quantization network by applying the Lipschitz constraint to the structural loss. In addition, we also for the first time take the robustness of the quantized NN into consideration and propose a non-sensitive perturbation loss function by introducing an extraneous term of spectral norm. The experiments are conducted on CIFAR-10 and ImageNet datasets with popular NN( such as MoblieNetV2, SqueezeNet, ResNet20, etc). The experimental results show that the proposed algorithm is more competitive under 2-bit-precision than the state-of-the-art quantization methods. Meanwhile, the experimental results also demonstrate that the proposed method is robust under the FGSM adversarial samples attack.
Open AccessRobustness-aware 2-bit quantization with real-time performance for neural network
Quantized neural network (NN) with a reduced bit precision is an effective solution to reduces the computational and memory resource requirements and plays a vital role in machine learning. However, it is still challenging to avoid the significant accuracy degradation due to its numerical approximation and lower redundancy. In this paper, a novel robustness-aware 2-bit quantization scheme is proposed for NN base on binary NN and generative adversarial network(GAN), witch improves the performance by enriching the information of binary NN, efficiently extract the structural information and considering the robustness of the quantized NN. Specifically, using shift addition operation to replace the multiply-accumulate in the quantization process witch can effectively speed the NN. Meanwhile, a structural loss between the original NN and quantized NN is proposed to such that the structural information of data is preserved after quantization. The structural information learned from NN not only plays an important role in improving the performance but also allows for further fine tuning of the quantization network by applying the Lipschitz constraint to the structural loss. In addition, we also for the first time take the robustness of the quantized NN into consideration and propose a non-sensitive perturbation loss function by introducing an extraneous term of spectral norm. The experiments are conducted on CIFAR-10 and ImageNet datasets with popular NN( such as MoblieNetV2, SqueezeNet, ResNet20, etc). The experimental results show that the proposed algorithm is more competitive under 2-bit-precision than the state-of-the-art quantization methods. Meanwhile, the experimental results also demonstrate that the proposed method is robust under the FGSM adversarial samples attack.
- Open Access
- 29 March 2018
- Published by IOP Publishing in IOP Conference Series: Materials Science and Engineering
- Vol. 322, 052032
A low delay transmission method of multi-channel video based on FPGA
- 29 March 2018
- Published by IOP Publishing in IOP Conference Series: Materials Science and Engineering
- Vol. 322, 052032
Open AccessA low delay transmission method of multi-channel video based on FPGA