Datasets:

Modalities:
Document
ArXiv:
DOI:
License:
anousias commited on
Commit
b7a6456
·
verified ·
1 Parent(s): b84e799

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +28 -5
README.md CHANGED
@@ -32,9 +32,13 @@ configs:
32
  path: 4096-unrolled_n/*test*.tar
33
  ---
34
 
 
35
  # COSOCO: Compromised Software Containers Image Dataset
36
 
37
- [Paper](https://huggingface.co/papers/2504.03238)
 
 
 
38
 
39
  COSOCO (Compromised Software Containers) is a synthetic dataset of 3364 images representing benign
40
  and malware-compromised software containers. Each image in the dataset represents a dockerized
@@ -44,8 +48,27 @@ A benign software container will have installed commonly used harmless packages
44
  a compromised software container, will have, among harmless benign tools and packages, its underlying
45
  file system affected by some activated malware instance. Each compromised instance is accompanied by
46
  a mask, i.e. a black and white image which marks the pixels that correspond to the files of the
47
- underlying system that have been altered by a malware.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
48
 
49
- COSOCO aims to support the identification of compromised software containers via the task of image
50
- classification task and the identification of compromised files and file system regions inside a
51
- container via the image segmentation task.
 
32
  path: 4096-unrolled_n/*test*.tar
33
  ---
34
 
35
+
36
  # COSOCO: Compromised Software Containers Image Dataset
37
 
38
+ - **Paper:** [Malware Detection in Docker Containers: An Image is Worth a Thousand Logs](https://huggingface.co/papers/2504.03238)
39
+ - **Dataset Documentation:** [COSOCO Dataset Documentation](./docs/COSOCO-dataset-readme-v1_0.pdf)
40
+
41
+ ## Dataset Description
42
 
43
  COSOCO (Compromised Software Containers) is a synthetic dataset of 3364 images representing benign
44
  and malware-compromised software containers. Each image in the dataset represents a dockerized
 
48
  a compromised software container, will have, among harmless benign tools and packages, its underlying
49
  file system affected by some activated malware instance. Each compromised instance is accompanied by
50
  a mask, i.e. a black and white image which marks the pixels that correspond to the files of the
51
+ underlying system that have been altered by a malware. COSOCO aims to support the identification of
52
+ compromised software containers via the task of image classification task and the identification of
53
+ compromised files and file system regions inside a container via the image segmentation task.
54
+
55
+ ## Citation
56
+
57
+ The users of this dataset are kindly asked to cite the following paper:
58
+
59
+ ```bibtex
60
+ @misc{
61
+ nousias2025malwaredetectiondockercontainers,
62
+ title={Malware Detection in Docker Containers: An Image is Worth a Thousand Logs},
63
+ author={Akis Nousias and Efklidis Katsaros and Evangelos Syrmos and Panagiotis Radoglou-Grammatikis and Thomas Lagkas and Vasileios Argyriou and Ioannis Moscholios and Evangelos Markakis and Sotirios Goudos and Panagiotis Sarigiannidis},
64
+ year={2025},
65
+ eprint={2504.03238},
66
+ archivePrefix={arXiv},
67
+ primaryClass={cs.CR},
68
+ url={https://arxiv.org/abs/2504.03238},
69
+ }
70
+ ```
71
+
72
+ ## Contact
73
 
74
+ Contact [Panagiotis Radoglou-Grammatikis](mailto:[email protected]) for questions or comments.