publications

2026

1. EAAI

   

   Text-based three-dimensional geometric person retrieval

   Fanzhi Jiang, Kexin Wang, Hanchi Ren, and 5 more authors

   Engineering Applications of Artificial Intelligence. Official article record available through ScienceDirect. Metadata cross-checked with Crossref and DBLP. , May 2026

   Abs DOI Bib HTML

   This work studies text-based retrieval for three-dimensional geometric person data, connecting natural-language person descriptions with 3D representations for person search. The paper appears in Engineering Applications of Artificial Intelligence and addresses a multimodal retrieval setting where text semantics must align with geometric person cues such as body shape, structure, and point-cloud-style spatial features.

   @article{jiang2026text3d,
     title = {{{Text-based three-dimensional geometric person retrieval}}},
     author = {Jiang, Fanzhi and Wang, Kexin and Ren, Hanchi and Li, Yiming and Zhang, Liumei and Hu, Yuanjiao and Xie, Xianghua and Yang, Su},
     journal = {Engineering Applications of Artificial Intelligence},
     volume = {171},
     pages = {114224},
     year = {2026},
     month = may,
     publisher = {Elsevier},
     doi = {10.1016/j.engappai.2026.114224},
     url = {https://doi.org/10.1016/j.engappai.2026.114224},
   }

2025

1. RoboticCC

   

   AAO-Grasp: Diffusion-Enhanced Object-Level Grasp All at Once

   Yiming Li, Hanchi Ren, Yue Yang, and 1 more author

   In 2025 International Conference on Robotic Computing and Communication (RoboticCC). Official IEEE record: 11391801. , Dec 2025

   Abs DOI Bib HTML

   AAO-Grasp targets 6-DoF robotic grasping in cluttered scenes. The work frames object-level grasp generation as an all-at-once process enhanced by diffusion modeling, aiming to produce coherent grasp hypotheses for objects rather than relying only on isolated local proposals.

   @inproceedings{li2025aaograsp,
     title = {{{AAO-Grasp}: Diffusion-Enhanced Object-Level Grasp All at Once}},
     author = {Li, Yiming and Ren, Hanchi and Yang, Yue and Xie, Xianghua},
     booktitle = {2025 International Conference on Robotic Computing and Communication (RoboticCC)},
     pages = {108--111},
     year = {2025},
     month = dec,
     publisher = {IEEE},
     doi = {10.1109/RoboticCC68732.2025.00030},
     url = {https://doi.org/10.1109/RoboticCC68732.2025.00030},
   }

2. RoboticCC

   

   Recognizing Robotic Components in Complex Environment

   Hanchi Ren, Yiming Li, Yunying Wang, and 7 more authors

   In 2025 International Conference on Robotic Computing and Communication (RoboticCC). Official IEEE record: 11391760. , Dec 2025

   Abs DOI Bib HTML

   This work focuses on robotic visual perception in cluttered and complex scenes, with an emphasis on automatic segmentation and recognition of robotic arm components. It supports downstream robotic interaction by separating robot structure from visually noisy environments where component boundaries and backgrounds are difficult to distinguish.

   @inproceedings{ren2025roboticcomponents,
     title = {Recognizing Robotic Components in Complex Environment},
     author = {Ren, Hanchi and Li, Yiming and Wang, Yunying and Guran, Nurhan Bulus and Hu, Chen and Yang, Yue and Hu, Yueyue and Hu, Junyan and Deng, Jingjing and Xie, Xianghua},
     booktitle = {2025 International Conference on Robotic Computing and Communication (RoboticCC)},
     pages = {92--99},
     year = {2025},
     month = dec,
     publisher = {IEEE},
     doi = {10.1109/RoboticCC68732.2025.00024},
     url = {https://doi.org/10.1109/RoboticCC68732.2025.00024},
   }

3. ICARM

   

   Triplane Grasping: Efficient 6-DoF Grasping with Single RGB Images

   Yiming Li, Hanchi Ren, Yue Yang, and 2 more authors

   In 2025 International Conference on Advanced Robotics and Mechatronics (ICARM). Preprint and method figures are available on arXiv. , Aug 2025

   Abs DOI arXiv Bib HTML PDF

   Triplane Grasping estimates 6-DoF grasps from a single RGB image by lifting visual evidence into a triplane-Gaussian scene representation. The method uses DINOv2 image features, point-cloud tokens, and triplane representation tokens to infer dense 3D structure before generating and filtering contact-based grasp candidates.

   @inproceedings{li2025triplane,
     title = {Triplane Grasping: Efficient 6-DoF Grasping with Single RGB Images},
     author = {Li, Yiming and Ren, Hanchi and Yang, Yue and Deng, Jingjing and Xie, Xianghua},
     booktitle = {2025 International Conference on Advanced Robotics and Mechatronics (ICARM)},
     pages = {1--7},
     year = {2025},
     month = aug,
     publisher = {IEEE},
     doi = {10.1109/ICARM65671.2025.11293550},
     url = {https://doi.org/10.1109/ICARM65671.2025.11293550},
   }

2024

1. COMCONF

   

   Gradients Stand-in for Defending Deep Leakage in Federated Learning

   Yi Hu, Hanchi Ren, Chen Hu, and 3 more authors

   In 2024 International Conference on Computing in Natural Sciences, Biomedicine and Engineering (COMCONF). Preprint and figures are available on arXiv; official implementation is in the AdaDefense repository. , Aug 2024

   Abs DOI arXiv Bib HTML PDF Code

   This paper introduces AdaDefense, a defense against gradient leakage in federated learning. Instead of transmitting raw gradients, the method sends gradient stand-ins based on lower-order moment adaptation, reducing the information exposed to reconstruction attacks while preserving the federated optimization workflow.

   @inproceedings{hu2024adadefense,
     title = {Gradients Stand-in for Defending Deep Leakage in Federated Learning},
     author = {Hu, Yi and Ren, Hanchi and Hu, Chen and Li, Yiming and Deng, Jingjing and Xie, Xianghua},
     booktitle = {2024 International Conference on Computing in Natural Sciences, Biomedicine and Engineering (COMCONF)},
     pages = {53--64},
     year = {2024},
     month = aug,
     publisher = {IEEE},
     doi = {10.1109/COMCONF63340.2024.00017},
     url = {https://doi.org/10.1109/COMCONF63340.2024.00017},
   }

2. ISBI

   

   CenterSAM: Fully Automatic Prompt for Dense Nucleus Segmentation

   Yiming Li, Hanchi Ren, Jingjing Deng, and 2 more authors

   In 2024 IEEE International Symposium on Biomedical Imaging (ISBI). Official code and visual results are available in the CenterSAM repository. Accepted manuscript hosted by Swansea Cronfa. , May 2024

   Abs DOI Bib HTML PDF Code

   CenterSAM is a fully automatic nucleus segmentation method that combines CenterNet-style object localization with the Segment Anything Model. A stacked hourglass detector predicts dense nucleus prompts such as centers and bounding boxes, which are passed to SAM to produce segmentation masks. The work evaluates the approach on 2018 Data Science Bowl, MoNuSeg, and TissueNet to test robustness across small, medium, and large biomedical image datasets.

   @inproceedings{li2024centersam,
     title = {{{CenterSAM}: Fully Automatic Prompt for Dense Nucleus Segmentation}},
     author = {Li, Yiming and Ren, Hanchi and Deng, Jingjing and Ma, Xiaoke and Xie, Xianghua},
     booktitle = {2024 IEEE International Symposium on Biomedical Imaging (ISBI)},
     pages = {1--5},
     year = {2024},
     month = may,
     publisher = {IEEE},
     doi = {10.1109/ISBI56570.2024.10635872},
     url = {https://doi.org/10.1109/ISBI56570.2024.10635872},
   }