Dataset & Benchmark

CAMotion

CAmouflaged Moving object detection

Paper Download Evaluation
GitHub

Introduction

Discovering camouflaged objects is a challenging task in computer vision due to the high similarity between camouflaged objects and their surroundings. While the problem of camouflaged object detection over sequential video frames has received increasing attention, the scale and diversity of existing video camouflaged object detection (VCOD) datasets are greatly limited, which hinders the deeper analysis and broader evaluation of recent deep learning-based algorithms with data-hungry training strategy. To break this bottleneck, we construct CAMotion, a high-quality benchmark that covers a wide range of species for camouflaged moving object detection in the wild. CAMotion comprises various sequences with multiple challenging attributes such as uncertain edge, occlusion, motion blur, and shape complexity, etc. The sequence annotation details and statistical distribution are presented from various perspectives, allowing CAMotion to provide in-depth analyses on the camouflaged object’s motion characteristics in different challenging scenarios.

CAMotion benchmark teaser figure
Figure 1. Examples from the CAMotion dataset with corresponding pixel-level annotations. Rows 1, 3, 5, and 7 show original images, while Rows 2, 4, 6, and 8 present the corresponding pixel-wise ground-truth annotations.

Statistics

149,319 Image Frames
6.51×MoCA-Mask
30,028 Annotated Frames
6.40×MoCA-Mask
152 Species
3.45×MoCA-Mask
Statistic comparison with other camouflage datasets
TABLE 1: Statistics of camouflage datasets. * indicates that the #Species is not reported in the original paper and is estimated by us.
Statistical comparison table of camouflage datasets
Figure 2: The scale and species comparison between existing camouflage datasets and CAMotion.
Figure 2: scale and species comparison between camouflage datasets and CAMotion
Figure 3: Scale distribution comparison of CAMotion and MoCA-Mask. Note that the reported ratio is defined as the proportion of foreground area relative to the entire image.
Figure 3: scale distribution comparison of CAMotion and MoCA-Mask

Dataset features

Figure 4 dataset feature visualization
Figure 4: Taxonomic structure of CAMotion. The inner ring illustrates the class taxonomy, and the outer ring shows the corresponding order taxonomy.
Figure 5 dataset feature visualization
Figure 5: An example of the hierarchy tree in CAMotion, illustrated with the Ray-finned Fish class.
Figure 6 dataset feature visualization
Figure 6: The attributes distribution of CAMotion in frame-level and sequence-level.

Tip: click to enlarge for details.

Figure 7 statistics for CAMotion dataset
Figure 7: Statistics for CAMotion dataset. (a) Object sizes distribution. (b) The distribution of video durations. (c) Global and local contrast distribution. (d) Motion statistics of the camouflaged objects.

Demo

Representative challenge categories in CAMotion. [TBD：与论文 Fig. 中类别命名对齐]

Low contrast
Low contrast
Occlusion
Occlusion
Motion ambiguity
Motion ambiguity
Small object
Small object

[TBD：按论文增加更多子类图块；Tab 的 data-filter 需与每个 figure 的 data-cat 一致]

Video teaser

[TBD：嵌入 MP4 / YouTube / Bilibili iframe]

Tasks & Evaluation

We report results under a unified protocol: define input (e.g., first-frame mask or box), allowed pretraining, and evaluation metrics. [TBD：1–2 句概括主实验设定，可引用论文 Sec.]

  • Metrics: mIoU, F-measure / S-measure, MAE [TBD：与终稿一致]
  • Protocol: [TBD：是否允许光流/外训数据等]
  • Test submission: [TBD：评测服务器提交流程一句话]

Table 2. Main benchmark results on CAMotion (placeholder). [TBD：从论文主表复制；可加「相对基线降幅」等一句 MOSE 式总结]

Method Backbone Setting mIoU MAE
Baseline-AREPLACESup.REPLACEREPLACE
Baseline-BREPLACESup.REPLACEREPLACE
OursREPLACESup.REPLACEREPLACE

Full leaderboard on Evaluation Server

Dataset

Dataset Download

The dataset is available for non-commercial research purposes only. Please use the following links.

[TBD：若与机构协议有关，粘贴许可声明原文或链接]

Hugging Face [TBD：填写实际 HF 数据集名] Google Drive [TBD：或改为百度网盘等]

[TBD：若仅一个源，删除多余卡片并把 href 改为 REPLACE_WITH_*]

Evaluation

Please submit results on the validation / test protocol as specified in the paper.

Codabench / Evaluation Server [TBD：平台名称与链接]

Data

CAMotion contains REPLACE videos and REPLACE high-quality masks for REPLACE objects across REPLACE categories. [TBD：与论文 Dataset 段数字一致]

  • Evaluation metrics follow standard segmentation practice (e.g., mIoU, boundary scores). [TBD]
  • For validation, first-frame annotations are released for the evaluated targets. [TBD：若与 MOSE 不同请改写]
  • Test-set policy: [TBD：公开/仅服务器/竞赛期开放等]

Data structure

train_valid.tar.gz
│
├── Annotations
│   ├── video_name_1
│   │   ├── 00000.png
│   │   └── ...
│   └── ...
│
└── JPEGImages
    ├── video_name_1
    │   ├── 00000.jpg
    │   └── ...
    └── ...

[TBD：按实际压缩包目录修改；与官方 README 完全一致]

Annotations only REPLACE size Splits & metadata REPLACE size Codebase GitHub
Usage notice. CAMotion is released for non-commercial research use. Please cite the paper and follow the license terms. [TBD：若采用 CC-BY-NC-SA 等，写明与 MOSE 相同或附许可证链接]

People

Siyuan Yao

[TBD：单位]

Hao Sun

[TBD：单位]

Hai Long

[TBD：单位]

Ruiqi Yu

[TBD：单位]

Jiehong Li

[TBD：单位]

Xiwei Jiang

[TBD：单位]

Yanzhao Su

[TBD：单位]

Wenqi Ren

[TBD：单位]

Xiaochun Cao

[TBD：单位 · 通讯作者可在此标注]

Citation

Please cite CAMotion if it helps your research.

@article{camotion,
  title   = {CAMotion: A High-Quality Dataset for Camouflaged Motion Object Detection in the Wild},
  author  = {Siyuan Yao and Hao Sun and Hai Long and Ruiqi Yu and Jiehong Li and Xiwei Jiang and Yanzhao Su and Wenqi Ren and Xiaochun Cao},
  journal = {Under review},
  year    = {2026}
}

[TBD：录用后替换 journal / 会议名、卷期、页码、DOI]

Related work

[TBD：可选：粘贴相关数据集或方法的 BibTeX，与 MOSE 页「Our related works」类似]

FAQ

How do I evaluate on the test split?

[TBD：作答]

Can I use external training data?

[TBD：作答]

Where to report issues?

[TBD：GitHub Issues / 邮箱]