Omni-Dimensional State Space Model-driven SAM for Pixel-level Anomaly Detection

Authors: Chao Huang, Qianyi Li, Jie Wen, Bob Zhang

IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results demonstrate that the proposed ODS-SAM outperforms state-of-the-art models on both industrial and medical image datasets.
Researcher Affiliation Academia Chao Huang1,2 , Qianyi Li3 , Jie Wen4 , Bob Zhang1 1University of Macau 2Sun Yat-sen University, Shenzhen Campus 3Ningbo University 4Harbin Institute of Technology, Shenzhen
Pseudocode No The paper describes the method using diagrams and textual explanations, but does not include any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statement about the release of source code or a link to a code repository.
Open Datasets Yes MVTec AD [Bergmann et al., 2020] is adopted to evaluate the proposed method for pixel-level anomaly detection in industrial scenes. The Mo Nu Seg dataset [Kumar et al., 2017] comprises 30 microscopy images from 7 organs in the training set, with annotations for 21,623 individual nuclei. The Gland segmentation (Gla S) challenge [Sirinukunwattana et al., 2017] encompasses 85 images for training and 80 images for testing. Moreover, four Polyp datasets (Kvasir-SEG [Jha et al., 2020], Clinic DB [Bernal et al., 2015], Colon DB [Tajbakhsh et al., 2015], and ETIS [Silva et al., 2014]) are adopted to evaluate the proposed method.
Dataset Splits Yes The Mo Nu Seg dataset [Kumar et al., 2017] comprises 30 microscopy images from 7 organs in the training set, with annotations for 21,623 individual nuclei. The Gland segmentation (Gla S) challenge [Sirinukunwattana et al., 2017] encompasses 85 images for training and 80 images for testing. The experiment setup is followed [Fan et al., 2020].
Hardware Specification Yes The batch size is 10 and trained on an NVIDIA A100 GPU.
Software Dependencies No The paper mentions using Adam optimizer but does not specify any software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes During the training, we adopted the Adam with an initial learning rate of 0.001 and a weight decay regularization parameter of 1e-5. The batch size is 10 and trained on an NVIDIA A100 GPU. The maximum number of epochs is 200. SAM pre-trained weights adopted in all experiments is based on Vi T-H.