---
title: "Armed People Detection Dataset"
description: "Explore Armed People Detection Dataset: 4,177 videos across 1,045 scenes, annotated for weapon types, carry positions, viewpoints, locations, and times of day"
url: "https://unidata.pro/datasets/armed-people-detection/"
date_modified: "2026-11-16T11:28:22+03:00"
language: "en-US"
---
This armed people dataset contains 4,177 videos in 1,045 scenes of a person carrying a weapon or similar objects. Each scene is filmed simultaneously from 4 sides (front, right, back, left) on 3 iPhones, indoors and outdoors, in the morning, day, evening, night. Annotations include weapon type (12 types), carry position (17 variants), location, time of day. Weapon detection dataset supports gun detection, people detection and threat-recognition models.

## List of Datasets

- [Graffiti Drawing Detection Video Dataset](https://unidata.pro/datasets/graffiti-drawing-detection-video/)
- [Fall Detection Dataset](https://unidata.pro/datasets/fall-detection/)
- [Weapon Detection Dataset](https://unidata.pro/datasets/weapon-detection/)
- [Fight Detection Video Dataset](https://unidata.pro/datasets/fight-detection-video/)

## Dataset Structure

### The Numbers Section

**Numbered list:** - **Number:** 4 177 — **Text:** Videos

### Tooltips Section

**Tooltip items:**

- **Name:** Computer Vision
- **Name:** Deep Learning
- **Name:** Public Safety
- **Name:** Object Detection
- **Name:** Smart Cities

### Dataset Information

**Table with data:**

| Characteristic | Data |
| --- | --- |
| Description | Videos of people carrying weapons or similar objects, filmed from multiple angles indoors and outdoors at various times of day with weapon type and carry position annotations, suitable for training threat-recognition models. |
| Data types | Video |
| Tasks | Public Safety, Computer Vision |
| Total number of files | 4,177 |
| Number of locations | 1,045 |
| Videos per scene | 4 in 1,042 scenes, 3 in 3 scenes |
| Labeling | Metadata (scene id, video id, camera angle, weapon type, weapon position, number of people, location, location id, time of day, device, resolution, fps, size bytes) |

**Media Slider:**

- **Video on Slayder:** <https://unidata.pro/wp-content/uploads/2026/11/armed-people-dataset.mp4>
- **Video on Slayder:** <https://unidata.pro/wp-content/uploads/2026/11/armed-people-dataset2.mp4>

**Link to the sample:** [Download sample](https://drive.google.com/drive/u/3/folders/1WFeQsqgCjX5CsscL_e2TCg_TH5B_tmcK)

### Technical Specifications

**Table with data:**

| Characteristic | Data |
| --- | --- |
| Video extension | MOV |
| Total size | about 170.8 GB |
| File size | 13.0–168.2 MB, median 35.7 MB, mean 40.9 MB |
| Resolution | 2160x1214 |
| Frame rate | 25 fps |
| Devices | iPhone 15 Pro, iPhone 12, iPhone 11 |

**Source and data collection methodology:** Source and collection methodology Data was captured via digital cameras in controlled, simulated environments.

### Statistics - Charts

**Charts with Titles:**

- **Shortcode:** [ays_chart id='107'] — **caption above the graph:** Camera Angle (by video)
- **Shortcode:** [ays_chart id='110'] — **caption above the graph:** Location
- **Shortcode:** [ays_chart id='111'] — **caption above the graph:** Time of Day (by scene)
- **Shortcode:** [ays_chart id='109'] — **caption above the graph:** Weapon Position (by scene)
- **Shortcode:** [ays_chart id='108'] — **caption above the graph:** Weapon Type (by scene)

### Dataset Use Cases - Slider

**Industry Cards:**

- **Industry:** Public Safety & Video Surveillance — **Title:** Real-time gun detection in CCTV feeds — **Text:** The armed people dataset helps train detection models that spot individuals carrying firearms in camera footage before violent incidents happen. Four camera angles and day-to-night recordings teach models to recognize weapons from any angle and in poor light, reducing false alarms in busy public spaces.
- **Industry:** Education & Commercial Security — **Title:** Weapons detection at building entrances — **Text:** Schools, offices and retail sites can use weapon detection datasets to build systems that flag guns and other weapons as people walk in. With 12 weapon types and 17 carry positions, the data covers many ways of holding objects, so security staff get earlier warnings and fewer missed risks.
- **Industry:** Transportation & Critical Infrastructure — **Title:** Threat monitoring in stations and airports — **Text:** Indoor and outdoor scenes make this gun detection dataset useful for transit hubs, stadiums and other crowded sites. Object detection models trained on 4,177 annotated videos can track armed people across morning, daytime, evening, and night conditions, supporting security teams that watch hundreds of cameras at once.
- **Industry:** AI Research & Model Benchmarking — **Title:** Testing people and weapon detection models — **Text:** Researchers can use this people detection dataset to compare how detection models perform across viewpoints, lighting, and carry styles. Annotations for weapon type, carry position, location, and time of day give clear data for error analysis, helping teams improve accuracy before deploying gun safety tools in real settings.

### Fact

**List of Questions:**

- **Question:** Can I request a sample of the armed people dataset before purchasing? — **Answer:** Yes. Unidata provides free samples of its datasets for trial and testing. A sample lets you check video quality, metadata structure, and compatibility with your object detection workflow before buying the full dataset.
- **Question:** What types of annotations are provided? — **Answer:** Each video is annotated with metadata: scene ID, video ID, camera angle, weapon type, weapon position, number of people, location, location ID, time of day, device, resolution, fps, size bytes. These labels support filtering, error analysis and training of gun detection and threat-recognition models. Frame-level annotations such as bounding boxes can be discussed with Unidata as a custom request.
- **Question:** How was the data collected? — **Answer:** The videos were recorded with three iPhone models (iPhone 15 Pro, iPhone 12 and iPhone 11) in controlled, simulated environments. Each scene was filmed simultaneously from four angles, giving detection models consistent multi-view data of the same person and weapon.
- **Question:** Can this gun detection dataset be used for real-time detection systems? — **Answer:** Yes. The multi-angle footage and varied lighting conditions help train object detection models that recognize firearms and other weapons in live video streams. This makes the dataset suitable for real-time gun detection in surveillance and security systems where early identification of risks is critical.
- **Question:** How are Unidata datasets licensed? — **Answer:** Unidata datasets follow a dual-licensing model. Free samples are provided for trial and testing, while full datasets are available exclusively through purchase.
- **Question:** How are Unidata datasets stored? — **Answer:** Unidata stores all datasets securely on AWS cloud infrastructure, which provides high availability and scalability. Storage and data management practices are aligned with ISO 27001 and ISO 27701, the international standards for information security and privacy management.
- **Question:** How long does it take to receive the dataset? — **Answer:** After you submit a request, the Unidata team contacts you to review the details and prepare the necessary documents. Once the agreement is signed and payment is made, the dataset is delivered within 3–10 days.
- **Question:** Is this a real-world dataset or synthetic data? — **Answer:** This is a real-world dataset, not synthetic or computer-generated data. All videos show real people filmed with iPhone cameras in staged scenarios, giving detection models natural motion, lighting, and background variation that synthetic data often lacks.

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