---
title: "2D Printed Photos Attacks and Replay Attacks Images Dataset"
description: "Large-scale 2D printed mask dataset with 4.1M+ images from 80,000+ individuals, designed for face anti-spoofing, liveness detection, and face recognition. It includes diverse printed attacks,…"
url: "https://unidata.pro/datasets/2d-printed-photos-attacks-and-replay-attacks-images-dataset/"
date_modified: "2025-11-19T16:43:26+03:00"
language: "en-US"
---
Large-scale 2D printed mask dataset with 4.1M+ images from 80,000+ individuals, designed for face anti-spoofing, liveness detection, and face recognition. It includes diverse printed attacks, replay attacks, face presentation attacks, and printed photos, with bounding box annotations, JSON labels, and metadata (age, gender, ethnicity) to support training data for anti-spoofing methods and detection algorithms.

The dataset covers multiple attack types, including screen replays, paper photos, color printouts, masks, tablets, and annotated mask samples, providing high-quality JPG images.

## Dataset Structure

### The Numbers Section

**Numbered list:**

- **Number:** 4,1M+ — **Text:** images
- **Number:** 80,000+ — **Text:** people

### Tooltips Section

**Tooltip items:**

- **Name:** Computer Vision
- **Name:** Machine learning
- **Name:** Security
- **Name:** Anti-spoofing
- **Name:** Liveness Detection

### Dataset Information

**Table with data:**

| Characteristic | Data |
| --- | --- |
| Description | Images of individuals with printed masks and replay attack |
| Data types | Image |
| Tasks | Anti-spoofing, Liveness Detection, Face Recognition |
| Number of image | 4,1M+ |
| Number of people | 80,000+ |
| Name of categories | archive(Faces displayed on screens), color(Color printouts of faces), easy(Masked faces on screens), frame, mask, print, faces on screen, all(Color printouts (generic)), bald, tablet, fake, Moscow newspapers, paper, mask annotated, paper photo |
| Labeling | Metadata (age, gender, ethnicity) |
| Gender | Male, Female |

**Media Slider:**

- **Image in the slider:** ![](https://unidata.pro/wp-content/uploads/2025/07/2d-printed-photos-attacks-and-replay-attacks-images-dataset0a-picture-2.webp)
- **Image in the slider:** ![](https://unidata.pro/wp-content/uploads/2025/07/2d-printed-photos-attacks-and-replay-attacks-images-dataset0a-picture-1.webp)

**Link to the sample:** [Download sample](https://drive.google.com/drive/folders/1Bjh9z4Gso470rhWaCFJVdeU6Jr-ZI5gj?usp=sharing)

### Technical Specifications

**Table with data:**

| Characteristic | Data |
| --- | --- |
| Image extension | JPG |
| Extension of labeling file | Bbox, json |

**Source and data collection methodology:** Source and collection methodology; Data was collected by a partner of Unidata.

### Dataset Use Cases - Slider

**Industry Cards:**

- **Industry:** Biometric Security — **Title:** Training Face Anti-Spoofing Models — **Text:** 2D Printed Mask dataset is well suited for developing face antispoofing models that distinguish genuine users from printed attacks and replay attacks. The combination of real faces and multiple presentation attacks provides valuable training data for improving spoofing detection and strengthening biometric recognition systems.
- **Industry:** Banking & Digital Identity — **Title:** Reducing Fraud During Remote Verification — **Text:** Digital identity platforms benefit from realistic attack samples when evaluating facial recognition pipelines. This Replay Attack Dataset helps assess how well authentication systems respond to printed photos, replay attacks, and other face spoofs, supporting more reliable remote onboarding and identity verification.
- **Industry:** AI & Computer Vision — **Title:** Evaluating Anti-Spoofing Algorithms — **Text:** Researchers developing detection algorithms can benchmark model performance using diverse presentation attacks captured in this anti-spoofing dataset. The variety of attack types and facial images enables consistent testing of anti-spoofing methods, image quality robustness, and recognition accuracy across different learning models.
- **Industry:** Access Control & Enterprise Security — **Title:** Enhancing Biometric Authentication — **Text:** Organizations building secure access control solutions rely on realistic attack scenarios to validate biometric systems. This dataset supports testing against printed attacks and replay attacks, helping improve face recognition performance while reducing the risk of successful presentation attacks in high-security environments.

### Fact

**FAQs Heading:** FAQs

**List of Questions:**

- **Question:** What annotations are included? — **Answer:** The dataset provides bounding box annotations, JSON labels, and demographic metadata including age, gender, and ethnicity. These annotations support attack detection, face recognition, detection algorithms, and machine learning workflows for biometric security.
- **Question:** Which attack types are represented in the dataset? — **Answer:** The dataset covers a wide range of presentation attacks, including screen replay attacks, printed photos, paper photos, color printouts, tablet displays, printed masks, fake faces, and annotated mask samples. This diversity helps AI models become more robust against multiple real-world spoofing techniques.
- **Question:** How was the dataset collected? — **Answer:** The dataset was collected by a Unidata partner specifically for AI development and biometric security applications. It contains high-quality images representing numerous printed attacks and replay attack scenarios suitable for training anti-spoofing methods.
- **Question:** Is this a real-world dataset or synthetic data? — **Answer:** This is a real-world image dataset, not synthetic data. It contains authentic images collected for biometric AI training and includes genuine examples of printed photos, screen replay attacks, and other presentation attack scenarios.
- **Question:** Can I download a sample of the dataset before purchasing? — **Answer:** Yes. Unidata provides a free sample of the Replay Attack Dataset so you can evaluate the image quality, annotation structure, metadata, and attack categories before purchasing the complete dataset. This allows you to verify compatibility with your anti-spoofing models and training workflow.
- **Question:** Can I request a custom anti-spoofing dataset? — **Answer:** Yes. Unidata can create custom anti-spoofing replay attack datasets tailored to your project requirements. We can customize attack types, demographic distribution, annotation formats, metadata, image specifications, and dataset size for your biometric security application.
- **Question:** How are Unidata datasets licensed? — **Answer:** Unidata datasets follow a dual-licensing model: free samples are available for evaluation and testing, while complete datasets are available exclusively through purchase. This allows organizations to validate the data before making a licensing decision.
- **Question:** Do Unidata datasets comply with GDPR and other privacy regulations? — **Answer:** Yes. All Unidata datasets are curated in compliance with GDPR and applicable data protection regulations. Data is collected from legally permissible sources to support ethical AI development and responsible biometric research.
- **Question:** How are Unidata datasets stored? — **Answer:** All datasets are securely stored on AWS cloud infrastructure with high availability and scalability. Storage and management practices follow ISO 27001 and ISO 27701 standards to ensure strong information security and privacy protection.

[Full list of this site's AI-readable pages](https://unidata.pro/llms.txt)