Yinghai Zhao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Yinghai Zhao has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Kuizhi Mei
State Key Laboratory of Human-Machine Hybrid Augmented Intelligence, Institute of Artificial Intelligence and Robotics, Xian Jiaotong University, Xian, China
5 shared publications - Xuyi Yu
State Key Laboratory of Human-Machine Hybrid Augmented Intelligence, Institute of Artificial Intelligence and Robotics, Xian Jiaotong University, Xian, China
4 shared publications - Huipeng Zhang
Institute of Artificial Intelligence and Robotics, State Key Laboratory of Human-Machine Hybrid Augmented Intelligence, Xian Jiaotong University, Xian, China
3 shared publications - Mengdao Xing
Xidian University
2 shared publications Ning Li
School of Chemistry and Chemical Engineering, Harbin Institute of Technology, Harbin 150001, People’s Republic of China
2 shared publications- Yanling Xu
School of Chemistry and Chemical Engineering, Harbin Institute of Technology, Harbin 150001, People's Republic of China
2 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 34 (11) , 3528-3539
Pipelined Microarchitecture for VTA With Circuit and Instruction Co-Optimization
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 34 (11) , 3528-3539
Pipelined Microarchitecture for VTA With Circuit and Instruction Co-Optimization
- 29 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. PP (10518215) , 1
Beyond Single Supervision: Multi-perspective Quantization-Aware Distillation for Semantic Segmentation
Efficient model compression is essential for deploying semantic segmentation models on edge devices. While quantization-aware training combined with knowledge distillation (KD) has emerged as a dominant paradigm, existing methods face a critical challenge: quantized teacher features provide deployment-compatible supervision but suffer from information loss, while full-precision features retain rich semantic details but exhibit representation mismatch with quantized students. Our analysis reveals that while single supervision sources each have inherent limitations, their combination produces superior results, suggesting the existence of more favorable intermediate representations in the feature space. To systematically explore this advantageous representation space, we propose Multi-perspective Feature Distillation (MFD). Our framework constructs multiple virtual quantized teachers by strategically perturbing the quantization parameters of a single full-precision teacher, generating diverse supervisory signals that bridge the gap between full-precision and quantized representations. This multi-perspective approach guides the student model toward learning robust features that maintain semantic richness while accommodating quantization constraints, all with minimal computational overhead. Comprehensive experiments show that MFD brings substantial improvements over existing distillation methods under 8-bit settings, while being efficient to train and readily integrable with existing KD frameworks. The source code will be available at https://github.com/Fateeeeee/MFD.
- 29 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. PP (10518215) , 1
Beyond Single Supervision: Multi-perspective Quantization-Aware Distillation for Semantic Segmentation
Efficient model compression is essential for deploying semantic segmentation models on edge devices. While quantization-aware training combined with knowledge distillation (KD) has emerged as a dominant paradigm, existing methods face a critical challenge: quantized teacher features provide deployment-compatible supervision but suffer from information loss, while full-precision features retain rich semantic details but exhibit representation mismatch with quantized students. Our analysis reveals that while single supervision sources each have inherent limitations, their combination produces superior results, suggesting the existence of more favorable intermediate representations in the feature space. To systematically explore this advantageous representation space, we propose Multi-perspective Feature Distillation (MFD). Our framework constructs multiple virtual quantized teachers by strategically perturbing the quantization parameters of a single full-precision teacher, generating diverse supervisory signals that bridge the gap between full-precision and quantized representations. This multi-perspective approach guides the student model toward learning robust features that maintain semantic richness while accommodating quantization constraints, all with minimal computational overhead. Comprehensive experiments show that MFD brings substantial improvements over existing distillation methods under 8-bit settings, while being efficient to train and readily integrable with existing KD frameworks. The source code will be available at https://github.com/Fateeeeee/MFD.
- 31 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (2) , 4858-4871
Cross-Head Relation Distillation for Edge-Friendly Semantic Segmentation in Transportation Systems
Semantic segmentation is a critical perception task for real-time scene understanding in consumer-grade edge devices, particularly in green transportation systems such as autonomous vehicles, smart dashcams, and in-vehicle ADAS units. However, the high computational cost of state-of-the-art models hinders their deployment in resource-constrained scenarios. Knowledge distillation offers a solution by transferring knowledge from a large teacher model to a compact student. Yet the prevalent multi-head architecture in modern teachers, which includes a main head together with auxiliary heads, points to a richer knowledge source than any single prediction. The interactions among these heads may encode valuable structural cues about class relationships and contextual dependencies, which are often overlooked in conventional distillation methods. Effectively distilling this cross-head organizational information, especially under architectural asymmetry, remains a significant challenge. To address this, we propose Cross-Head Relation Distillation (CHRD), a framework that explicitly models and transfers the structured knowledge derived from the teacher’s dual-head system. CHRD introduces cross-head consistency imitation to transfer a confidence-weighted consensus and cross-head relation imitation to capture class-level affinities. To address the structural asymmetry between teacher and student, we design a shared auxiliary head mechanism, enabling any student to leverage this cross-head knowledge. Furthermore, a loss normalization strategy ensures stable, progressive learning. Our method achieves competitive performance across multiple benchmarks, setting a new state-of-the-art on ADE20K and attaining highly competitive results on Cityscapes and CamVid, without relying on intermediate feature alignment. Code will be available at https://github.com/Fateeeeee/CHRD.
- 31 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (2) , 4858-4871
Cross-Head Relation Distillation for Edge-Friendly Semantic Segmentation in Transportation Systems
Semantic segmentation is a critical perception task for real-time scene understanding in consumer-grade edge devices, particularly in green transportation systems such as autonomous vehicles, smart dashcams, and in-vehicle ADAS units. However, the high computational cost of state-of-the-art models hinders their deployment in resource-constrained scenarios. Knowledge distillation offers a solution by transferring knowledge from a large teacher model to a compact student. Yet the prevalent multi-head architecture in modern teachers, which includes a main head together with auxiliary heads, points to a richer knowledge source than any single prediction. The interactions among these heads may encode valuable structural cues about class relationships and contextual dependencies, which are often overlooked in conventional distillation methods. Effectively distilling this cross-head organizational information, especially under architectural asymmetry, remains a significant challenge. To address this, we propose Cross-Head Relation Distillation (CHRD), a framework that explicitly models and transfers the structured knowledge derived from the teacher’s dual-head system. CHRD introduces cross-head consistency imitation to transfer a confidence-weighted consensus and cross-head relation imitation to capture class-level affinities. To address the structural asymmetry between teacher and student, we design a shared auxiliary head mechanism, enabling any student to leverage this cross-head knowledge. Furthermore, a loss normalization strategy ensures stable, progressive learning. Our method achieves competitive performance across multiple benchmarks, setting a new state-of-the-art on ADE20K and attaining highly competitive results on Cityscapes and CamVid, without relying on intermediate feature alignment. Code will be available at https://github.com/Fateeeeee/CHRD.
- 30 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (12) , 27181-27196
Toward Compact Representations: A Region-Based Logit Distillation Framework for Edge-Friendly Semantic Segmentation
Semantic segmentation enables fine-grained scene understanding for the Internet of Things (IoT) applications such as intelligent transportation and UAV-based monitoring. However, high-resolution segmentation models impose substantial memory and computational demands, making them impractical for resource-constrained IoT edge devices. Knowledge distillation (KD) transfers knowledge from a large teacher model to a compact student model. Yet, conventional logit distillation in segmentation suffers from two limitations: 1) dense pixel-wise alignment introduces spatial redundancy and 2) erroneous teacher predictions propagate to the student, degrading reliability. To address these issues, this article proposes a region-based logit distillation (RLD) framework. Instead of distilling all logits, RLD summarizes logit distributions within categories to reduce redundancy. To avoid interference from incorrect logits, we isolate their influence using label guidance and then summarize logits for a batch of samples via intraclass similarity. Teacher-guided singular value decomposition (SVD) facilitates logit dimensionality reduction. RLD retains only a small set of principal logits, for example, 19 for Cityscapes, achieving performance comparable to dense distillation. Combined with classic distillation baselines, RLD further achieves state-of-the-art results: 77.80% mIoU on Cityscapes and 69.90% on CamVid. Deployed on an NVIDIA Jetson Xavier NX, our distilled model achieves 46.78 FPS at 14.05 W in FP16 and 85.12 FPS at 7.73 W with INT8 quantization, demonstrating real-time capability and high energy efficiency for edge IoT applications.
- 30 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (12) , 27181-27196
Toward Compact Representations: A Region-Based Logit Distillation Framework for Edge-Friendly Semantic Segmentation
Semantic segmentation enables fine-grained scene understanding for the Internet of Things (IoT) applications such as intelligent transportation and UAV-based monitoring. However, high-resolution segmentation models impose substantial memory and computational demands, making them impractical for resource-constrained IoT edge devices. Knowledge distillation (KD) transfers knowledge from a large teacher model to a compact student model. Yet, conventional logit distillation in segmentation suffers from two limitations: 1) dense pixel-wise alignment introduces spatial redundancy and 2) erroneous teacher predictions propagate to the student, degrading reliability. To address these issues, this article proposes a region-based logit distillation (RLD) framework. Instead of distilling all logits, RLD summarizes logit distributions within categories to reduce redundancy. To avoid interference from incorrect logits, we isolate their influence using label guidance and then summarize logits for a batch of samples via intraclass similarity. Teacher-guided singular value decomposition (SVD) facilitates logit dimensionality reduction. RLD retains only a small set of principal logits, for example, 19 for Cityscapes, achieving performance comparable to dense distillation. Combined with classic distillation baselines, RLD further achieves state-of-the-art results: 77.80% mIoU on Cityscapes and 69.90% on CamVid. Deployed on an NVIDIA Jetson Xavier NX, our distilled model achieves 46.78 FPS at 14.05 W in FP16 and 85.12 FPS at 7.73 W with INT8 quantization, demonstrating real-time capability and high energy efficiency for edge IoT applications.
- 30 January 2026
- Published by Elsevier in Neural Networks
- Vol. 199, 108663
Efficient semantic segmentation via logit-guided feature distillation
- 30 January 2026
- Published by Elsevier in Neural Networks
- Vol. 199, 108663
Efficient semantic segmentation via logit-guided feature distillation
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64, 1002214
An Ensemble Recognition Model Based on Cross-Modal Fusion of HRRP and Infrared Image
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64, 1002214
An Ensemble Recognition Model Based on Cross-Modal Fusion of HRRP and Infrared Image
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Aerospace and Electronic Systems
- Vol. 62, 14900-14917
Robust Perception of Compound Jamming via Pulse-Compression-Domain Blind Source Separation
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Aerospace and Electronic Systems
- Vol. 62, 14900-14917
Robust Perception of Compound Jamming via Pulse-Compression-Domain Blind Source Separation
- Top Cited Papers
- 13 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-15
Multilabel Deep Learning-Based Lightweight Radar Compound Jamming Recognition Method
With the rapid development of electronic countermeasure technology, many active jamming compound scenes pose severe challenges to traditional radar, synthetic aperture radar (SAR), and other detection technologies. The accurate monitoring and recognition of individual jamming types contained in the complex electromagnetic environment can provide valuable prior information for radar countermeasures. However, existing jamming recognition algorithms suffer from huge models, fewer recognizable jamming types, and weak robustness, which is difficult to apply effectively to the resource-constrained airborne pulse signal real-time analysis instruments. This paper proposes a multi-label learning-based lightweight compound jamming recognition algorithm to solve these problems, including three key steps. First, the proposed method performs de-chirp, time-frequency transformation, and grayscale compression preprocessing for radar echoes. Then, an efficient hybrid attention (EHA) mechanism is designed and combined with ShuffleNet v2 to construct a recognition model. Finally, we generate independent multi-label discriminant thresholds based on dual evaluation metrics and a genetic algorithm to improve the recognition effect. The experiment shows that the floating-point operation (FLOPs) of the proposed method is only 0.11% ~ 57.19% of the existing jamming recognition methods, the overall recognition accuracy of the measured jamming data is 92.25%, higher than the existing methods of 7.37% ~ 16.73%, and has strong robustness to the fluctuation of radar waveform parameters.
- 13 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-15
Top Cited PapersMultilabel Deep Learning-Based Lightweight Radar Compound Jamming Recognition Method
With the rapid development of electronic countermeasure technology, many active jamming compound scenes pose severe challenges to traditional radar, synthetic aperture radar (SAR), and other detection technologies. The accurate monitoring and recognition of individual jamming types contained in the complex electromagnetic environment can provide valuable prior information for radar countermeasures. However, existing jamming recognition algorithms suffer from huge models, fewer recognizable jamming types, and weak robustness, which is difficult to apply effectively to the resource-constrained airborne pulse signal real-time analysis instruments. This paper proposes a multi-label learning-based lightweight compound jamming recognition algorithm to solve these problems, including three key steps. First, the proposed method performs de-chirp, time-frequency transformation, and grayscale compression preprocessing for radar echoes. Then, an efficient hybrid attention (EHA) mechanism is designed and combined with ShuffleNet v2 to construct a recognition model. Finally, we generate independent multi-label discriminant thresholds based on dual evaluation metrics and a genetic algorithm to improve the recognition effect. The experiment shows that the floating-point operation (FLOPs) of the proposed method is only 0.11% ~ 57.19% of the existing jamming recognition methods, the overall recognition accuracy of the measured jamming data is 92.25%, higher than the existing methods of 7.37% ~ 16.73%, and has strong robustness to the fluctuation of radar waveform parameters.
- 1 September 2022
- Published by The Electrochemical Society in Journal of the Electrochemical Society
- Vol. 169 (9) , 092509
Numerical Simulation for Improving Coating Homogeneity of High-Speed Tin Plating System
The influence of electrodeposition coating uniformity is complex; in essence, coating thickness depends on the potential distribution on the cathode surface, which is jointly affected by geometric and electrochemical factors. In a realistic high-speed tin plating production process, the edge effect has always been a vital problem to the homogeneity of electrodeposition coatings. Due to the complexity of production equipment and the high costs of spot experiments, modeling and simulation have become promising ways to study how to improve the uniformity of coatings. In this paper, we proposed reasonable model simplification strategies in the high-speed tin electrodeposition process and studied the influence of anode arrangement and auxiliary equipment on the uniformity of tin plating coating based on COMSOL Multiphysics. The results showed that changing the relative positions of cathode and anode cannot make the coating evenly distributed while adding specific auxiliary devices at appropriate locations would helpfully solve the uneven coating problem caused by edge effects. The research not only aims at a high-speed tin plating system but also has a great reference with any plate-type cathode electrodeposition for improving the homogeneity of coatings.
- 1 September 2022
- Published by The Electrochemical Society in Journal of the Electrochemical Society
- Vol. 169 (9) , 092509
Numerical Simulation for Improving Coating Homogeneity of High-Speed Tin Plating System
The influence of electrodeposition coating uniformity is complex; in essence, coating thickness depends on the potential distribution on the cathode surface, which is jointly affected by geometric and electrochemical factors. In a realistic high-speed tin plating production process, the edge effect has always been a vital problem to the homogeneity of electrodeposition coatings. Due to the complexity of production equipment and the high costs of spot experiments, modeling and simulation have become promising ways to study how to improve the uniformity of coatings. In this paper, we proposed reasonable model simplification strategies in the high-speed tin electrodeposition process and studied the influence of anode arrangement and auxiliary equipment on the uniformity of tin plating coating based on COMSOL Multiphysics. The results showed that changing the relative positions of cathode and anode cannot make the coating evenly distributed while adding specific auxiliary devices at appropriate locations would helpfully solve the uneven coating problem caused by edge effects. The research not only aims at a high-speed tin plating system but also has a great reference with any plate-type cathode electrodeposition for improving the homogeneity of coatings.
- 1 August 2022
- Published by The Electrochemical Society in Journal of the Electrochemical Society
- Vol. 169 (8) , 082501
Passivation and Chemical Conversion Combined Multi-Elements Coating on Low Sn-Coated Steel for Corrosion Protection
In this work, the tinplate with 0.5 g·m −2 tin coating mass was prepared in order to improve the economic benefit for food packaging field. We studied this low Sn-coated steel surface state and developed a process combined Mo-Mn-Al-P coating based on its special structure. By Taguchi techniques optimization, this chromium-free multi-elements coating has excellent corrosion resistance and adhesion with organic coatings, would be a hopefully replacing traditional chromate passivation treatment on tinplate with low tin coating mass.
- 1 August 2022
- Published by The Electrochemical Society in Journal of the Electrochemical Society
- Vol. 169 (8) , 082501
Passivation and Chemical Conversion Combined Multi-Elements Coating on Low Sn-Coated Steel for Corrosion Protection
In this work, the tinplate with 0.5 g·m −2 tin coating mass was prepared in order to improve the economic benefit for food packaging field. We studied this low Sn-coated steel surface state and developed a process combined Mo-Mn-Al-P coating based on its special structure. By Taguchi techniques optimization, this chromium-free multi-elements coating has excellent corrosion resistance and adhesion with organic coatings, would be a hopefully replacing traditional chromate passivation treatment on tinplate with low tin coating mass.