ZLZeyuan Li

Zeyuan Li

*This is an algorithmically generated author record based on publicly available information.

Zeyuan Li has a total of 21 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Spectral Representation-Enhanced Multitask SAM for Cropland Parcel Delineation

    by Aoran Hu, Siyao Cheng, Ran Meng, Zeyuan Li, Jie Liu

    Accurate cropland parcel delineation (CLPD) from remote sensing imagery is crucial for fine-scale agricultural monitoring. The recent Segment Anything Model (SAM), as a foundational vision model, demonstrates remarkable generalization capabilities, presenting significant potential for CLPD. However, SAM’s design for RGB natural imagery limits its ability to model multispectral information effectively. To address these limitations, we propose a spectrally enhanced multitask segmentation framework built upon SAM for CLPD. Our approach enhances SAM in two key aspects: First, to better leverage multispectral data, we introduce a decoupled learning strategy designed to improve SAM’s spectral representation. It is implemented through two components: learnable channel tokens encoding band-specific spectral identities and a lightweight channel-attention module modeling inter-band relationships. Second, to enhance decoding robustness under high intra-parcel spectral variability, we introduce a feature-structure regularization loss that enforces intra-class compactness and inter-class separation, along with an auxiliary instance-guided decoding branch that captures fine-grained sub-parcel patterns. These patterns are integrated into the main segmentation stream via a token-based task interaction module. Extensive experiments on two public datasets demonstrate consistent improvements of the proposed method. On the Sen4AgriNet dataset, it achieves a pixel-level F1-score of 89.36% and reduces the object-level Global Total Classification Error by 0.97% compared with existing methods. Ablation studies further validate the contribution of each component and confirm the effectiveness of our decoupling strategy. The source code will be made available at: https://github.com/halfhalfer/MSSAM.

  • Open Access

    Blockchain-Enabled Identity Based Authentication Scheme for Cellular Connected Drones

    by Yu Su, Zeyuan Li, Yufei Zhang, Xun Gui, Xue Deng, Jun Fu

    The proliferation of drones across precision agriculture, disaster response operations, and delivery services has accentuated the critical need for secure communication frameworks. Due to the limited computational capabilities of drones and the fragility of real-time wireless communication networks, the cellular connected drones confront mounting cybersecurity threats. Traditional authentication mechanisms, such as public-key infrastructure-based authentication, and identity-based authentication, are centralized and have high computational costs, which may result in single point of failure. To address these issues, this paper proposes a blockchain-enabled authentication and key agreement scheme for cellular-connected drones. Leveraging identity-based cryptography (IBC) and the Message Queuing Telemetry Transport (MQTT), the scheme flow is optimized to reduce the communication rounds in the authentication. By integrating MQTT brokers with the blockchain, it enables drones to authenticate through any network node, thereby enhancing system scalability and availability. Additionally, cryptographic performance is optimized via precompiled smart contracts, enabling efficient execution of complex operations. Comprehensive experimental evaluations validate the performance, scalability, robustness, and resource efficiency of the proposed scheme, and show that the system delivers near-linear scalability and accelerated on-chain verification.


  • Open Access

    Incremental Object Detection based on YOLO v5 and EWC Models

    by Zeyuan Li, Xinyan Liu, Cheng Zhang, Shunming Li, Zhiyu Xi, Wei Deng

    Current object detection models become increasingly sophisticated under the challenges of open domain. It is strongly needed to continuously learn new unseen images to achieving incremental object detection capacity. The existing methods mainly rely on two-stage object detection models which have shortcomings of time consuming and not real-time detection. In this paper, we propose a new incremental object detection model combining the real-time object detection model named YOLO with the incremental learning model EWC (Elastic Weight Consolidation), which is a posterior information-based model balancing the importance of new and old parameters. Our model modifies the loss function of YOLO and adds the EWC model as a regularization term. Experimental results validate the validity of our model on the Pascal VOC2012 datasets.


  • Open Access

    Crowd-aware Black Pig Detection for Low Illumination

    by Cheng Zhang, Shunming Li, Zeyuan Li, Na Wu, Qian Miao, Shijia Li

    Pig detection is vital to pig farms since it is the basis for counting, weight estimation, and behavior recognition functions. Existing methods focused on white pig detection instead of black pigs because the contrast between white pigs and the background is more evident than black pigs, making black pig detection more challenging. Furthermore, pig farms often suffer from insufficient light, and pigs tend to be stacked together, which makes it hard to detect pigs accurately. To this end, we propose a black pig detection method, based on YOLOV4, robust to both crowd scenes and low-illumination conditions. The method consists of the following parts: 1) generate domain adaptation dataset (DAD) based on style-transfer to optimize the original data distribution, therefore improving the performance of the method for low-illuminated conditions; 2) propose a crowd-aware module (CAM), adapted to YOLOV4 backbone architecture, to generate crowd density maps; 3) develop an adaptive attention module (AAM) to fuse YOLOV4 backbone features with corresponding crowd density map allowing the method robust to pigs in the crowd. The experimental results confirm the feasibility of this method. The mAP value in fully-lighting and poorly-lighting increased to 88.95% and 84.74%, respectively.