Awesome

<p align=center>Awesome Concealed Object Segmentation </p>

Contents

• Latest Works Recommended
• Citation
• Papers
  • Camouflaged Object Detection(COD)
    • Image-level COD
      • Normal-task setting
        • RGB COD
        • RGB-D COD
      • Novel-task setting
    • Video-level COD(VCOD)
      • Two-step framework
      • End-to-end framework
  • Transparent Object Segmentation(TOS)
  • Surface Defect Detection(SDD)
  • Polyp Segmentation(PS)
  • Lung Nodule Detection(LND)
• Datasets
  • COD
  • TOS
  • SDD
  • PS
  • LND
• Related Surveys Recommended
• Reference

<span id = "latest-works-recommended">Latest Works Recommended</span>

Leveraging Hallucinations to Reduce Manual Prompt Dependency in Promptable Segmentation ProMaC<br/>Jian Hu, Jiayi Lin, Junchi Yan, Shaogang Gong<br/>NeurIPS, 2024. [Paper] [Code]<br/>Sep. 2024

A Survey of Camouflaged Object Detection and Beyond <br/>Fengyang Xiao, Sujie Hu, Yuqi Shen, Chengyu Fang, Jinfa Huang, Chunming He, Longxiang Tang, Ziyun Yang, Xiu Li<br/>arxiv, 2024. [Paper] [Code]<br/>Aug. 2024

<span id = "Citation">Citation</span>

If you find our work useful in your research, please consider citing:

@article{2024CODsurvey,
  title={A Survey of Camouflaged Object Detection and Beyond},
  author={Xiao, Fengyang and Hu, Sujie and Shen, Yuqi and Fang, Chengyu and Huang, Jinfa and He, Chunming and Tang, Longxiang and Yang, Ziyun and Li, Xiu},
  journal={arXiv preprint arXiv:2408.14562},
  year={2024}
}

<span id = "papers">Papers</span>

<span id = "1-camouflaged-object-detectioncod">1. Camouflaged Object Detection(COD)</span>

<span id = "11-Image-level-COD">1.1 Image-level COD</span>

<span id = "111-Normal-task-setting">1.1.1 Normal-task setting</span>

Image-level COD methods in normal-task setting use a variety of different strategies, mainly in the following five categories. And we include tags related to core techniques for each item.

• Multi-scale-context (MSC)
• Mechanism-simulation (MS)
• Multi-source information fusion (MSI)
• Multi-task learning (MTL)
• Joint-SOD (JS)

<span id = "RGB-COD">1.1.1.1 RGB COD</span>

<span id = "RGB-D-COD">1.1.1.2 RGB-D COD</span>

<span id = "112-Novel-task-setting">1.1.2 Novel-task setting</span>

• Unsupervised camouflaged object segmentation (UCOS)

• Collaborative camouflaged object detection (CoCOD)

• Referring camouflaged object detection (RefCOD)

• Open-vocabulary camouflaged object segmentation (OVCOS)

<span id = "12-Video-level-COD">1.2 Video-level COD(VCOD)</span>

<span id = "121-Novel-task-setting">1.2.1 Two-step framework</span>

<span id = "122-End-to-end-framework">1.2.2 End-to-end framework</span>

<span id = "2-transparent-object-segmentationtos">2. Transparent Object Segmentation(TOS)</span>

<span id = "p_SDD">3. Surface Defect Detection(SDD)</span>

<span id = "4-polyp-segmentationps">4. Polyp Segmentation(PS)</span>

<span id = "5-lung-nodule-detectionlnd">5. Lung Nodule Detection(LND)</span>

<span id = "Datasets">Datasets</span>

<span id = "1-camouflaged-object-detectioncod-1">1. Camouflaged Object Detection(COD)</span>

The basic information of both image-level and video-level COD datasets.

• Level: data type of dataset, i.e., Image(I) and Video(V).
• Train/Test: number of samples for training/testing, e.g., images for image dataset or frames for video dataset.
• Resolution: image or frame resolution, some may be a range from the smallest area to the largest area.
• Objects
  • object categories of datasets.
  • **N. Cam. **: whether collecting non-camouflaged samples.
• Labels
  • Cls.: whether providing classification labels for Camouflaged Objects Classification(COCls). If so, the number of categories are provided.
  • Fix.: whether providing fixation annotation for Camouflaged Objects Localization(COL).
  • B. Box: whether providing bounding box labels.
  • Obj.: whether providing object-level segmentation masks.
  • Ins.: whether providing instance-level segmentation masks for Camouflaged Instance Segmentation (CIS).
  • Ran.: whether providing ranking labels for Camouflaged Instance Rank (CIR).
  • Scr.: whether providing weakly-supervised labels in scribbled form.
  • Gro.: whether providing corresponding category annotation within group images for Collaborative Camouflaged Object Detection (CoCOD).
  • Ref.: whether providing referring images for Referring Camouflaged Object Detection (RefCOD).
  • Uns.: whether providing unseen classes for Open-Vocabulary Camouflaged Object Segmentation (OVCOS).

<span id = "2-transparent-object-segmentationtos-1">2. Transparent Object Segmentation(TOS)</span>

<span id = "3-surface-defect-detectionsdd-1">3. Surface Defect Detection(SDD)</span>

<span id = "4-polyp-segmentationps-1">4. Polyp Segmentation(PS)</span>

<span id = "5-lung-nodule-detectionlnd-1">5. Lung Nodule Detection(LND)</span>

<span id = "related-surveys-recommended">Related Surveys Recommended</span>

A Survey of Camouflaged Object Detection and Beyond COD <br/>arxiv 2024. [Paper] [Code]<br/>Aug. 2024

Advances in deep concealed scene understanding COD CDS2K dataset<br/>VI 2023. [Paper] [Code]<br/>Aug. 2023

A systematic review of image-level camouflaged object detection with deep learning COD<br/>Neucom 2024. [Paper]<br/>Jan. 2024

Rethinking Camouflaged Object Detection: Models and Datasets COD<br/> TCSVT 2021. [Paper]<br/>Nov. 2021

Advancements in Optimization Algorithms for Lung Nodule Detection and Classification: A Review LND<br/>ICOTL 2023. [Paper]<br/>Dec. 2023

Detection of Lung Nodules using Convolution Neural Network: A Review LND<br/>ICIRCA 2020. [Paper]<br/>Jul. 2020

<span id = "reference">Reference</span>

SINet-V2/AWESOME_COD_LIST.md

visionxiang/awesome-camouflaged-object-detection