LSKNet: A foundation lightweight backbone for remote sensing (IJCV)

Yuxuan Li, Xiang Li#, Yimian Dai, Qibin Hou, Li Liu, Yongxiang Liu, Ming-Ming Cheng, Jian Yang#

[Paper] [BibTex] [Demo Video] [Code]

LSKNet can dynamically adjust its large spatial receptive field to better model the ranging context of various categories of objects in remote sensing scenarios.

APF-GAN:Exploring asymmetric pre-training and fine-tuning strategy for conditional generativeadversarial network (CVMJ)

Yuxuan Li, Lingfeng Yang, Xiang Li#

[Paper] [Code]

The APF-GAN model improves GAN-based image segmentation, surpassing GauGAN and winning the Second Jittor AI Challenge.

SM3Det: A Unified Model for Multi-Modal Remote Sensing Object Detection (AAAI 2026 Oral)

Yuxuan Li, Xiang Li#, Yunheng Li, Yicheng Zhang, Yimian Dai, Qibin Hou, Ming-Ming Cheng and Jian Yang#

[Paper] [BibTex] [Code] [知乎专栏]

This paper introduces SM3Det, a unified remote sensing model designed to handle multi-modal, multi-task object detection by leveraging a sparse Mixture-of-Experts backbone and dynamic optimization.

SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection (NeurIPS 2024 Spotlight)

Yuxuan Li, Xiang Li#, Weijie Li, Qibin Hou, Li Liu, Ming-Ming Cheng, Jian Yang#

[Paper] [BibTex] [Code] [知乎专栏]

SARDet-100K, the first large-scale multi-class SAR object detection dataset, along with the proposed MSFA pretraining framework, addresses domain and model disparities, significantly improving SAR object detection performance and advancing the field.

Large Selective Kernel Network for Remote Sensing Object Detection (ICCV 2023)

Yuxuan Li, Qibin Hou, Zhaohui Zheng, Ming-Ming Cheng, Jian Yang#, Xiang Li#

[Paper] [BibTex] [Demo Video] [Report/Forum] [Code] [知乎专栏]

LSKNet can dynamically adjust its large spatial receptive field to better model the ranging context of various categories of objects in remote sensing scenarios.

Spatial group-wise enhance: Enhancing semantic feature learning in CNN (ACCV 2022)

Yuxuan Li, Xiang Li#, Jian Yang

[Paper] [BibTex] [[Code]] [知乎专栏]

The Spatial Group-wise Enhance (SGE) module improves CNN performance by generating accurate spatial attention masks through local-global similarity within semantic groups, achieving notable accuracy gains on ImageNet and COCO tasks with minimal overhead.