Starred repositories
Patchwork++: Fast and robust ground segmentation method for 3D LiDAR scans. @ IROS'22
Unofficial implementation of "MapCleaner" a method for removing moving objects from point cloud maps.
GLIM: versatile and extensible point cloud-based 3D localization and mapping framework
利用UcoSLAM对ORB特征和Marker位姿鲁棒估计的特性，改进FAST-LIVO2框架因为激光特征退化导致的偏移问题。更改部分是EKF的VIO部分，将光度误差形式的残差格式改为由Uco提供的pose graph的残差格式，并且放在统一的真实世界尺度下，全局共用一套state状态量，提供了数值稳定的EKF紧耦合修改。充分发挥了marker对Lidar建图提供的效果提升。
Official code for the paper: "ReplayCAD: Generative Diffusion Replay for Continual Anomaly Detection" (IJCAI 2025)
Official implementation of "Stable Diffusion Zero-Shot Anomaly Detection with Segment Anything".
"Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancement" (ICCV 2023) & (NTIRE 2024 Runner-Up)
Official implementation of paper FiLo: Zero-Shot Anomaly Detection by Fine-Grained Description and High-Quality Localization (ACM MM 2024).
[CVPR 2025] official implementation of “Exploring Intrinsic Normal Prototypes within a Single Image for Universal Anomaly Detection”
Caption free adapter that maps DINOv3 image embeddings into CLIP space so you can do zero-shot text -> image or image -> text with CLIP’s text tower
SegDINO: An Efficient Design for Medical and Natural Image Segmentation with DINO-V3
Reference PyTorch implementation and models for DINOv3
Official implementation for AnomalyCLIP (ICLR 2024)
Official implementation of "Segment Any Anomaly without Training via Hybrid Prompt Regularization (SAA+)".
[TII 2025] Official Implementation and Dataset Release for "Center-aware Residual Anomaly Synthesis for Multi-class Industrial Anomaly Detection"
Context-Guided Prompt Learning and Attention Refinement for Zero-Shot Anomaly Detection
We proposes IAD-R1, a universal post-training framework that enhances Vision-Language Models for industrial anomaly detection through a two-stage training strategy.
[ArXiv 2025] Official Implementation for "CoPS: Conditional Prompt Synthesis for Zero-Shot Anomaly Detection"
[CVPR 2025] Official Implementation of "Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection". The first multi-class UAD model that can compete with single-class SOTAs
(CVPR2025) the code of "Bayesian Prompt Flow Learning for Zero-Shot Anomaly Detection"
(ICCV 2025) DictAS: A Framework for Class-Generalizable Few-Shot Anomaly Segmentation via Dictionary Lookup
[ICML2025] Official Implementation of CostFilter-AD: Enhancing Anomaly Detection through Matching Cost Filtering
Official implementation of "Unseen Visual Anomaly Generation" (CVPR 2025)
[ICCV 2025] SALAD -- Semantics-Aware Logical Anomaly Detection
[ICCV2025] SeaS: Few-shot Industrial Anomaly Image Generation with Separation and Sharing Fine-tuning. Paper is available at https://arxiv.org/abs/2410.14987