--- title: "Fight Detection Video Dataset" description: "This fight detection dataset contains videos of simulated physical altercations recorded in controlled environments, captured from static and moving surveillance camera views at up to…" url: "https://unidata.pro/datasets/fight-detection-video/" date_modified: "2026-03-03T14:37:28+03:00" language: "en-US" --- This fight detection dataset contains videos of simulated physical altercations recorded in controlled environments, captured from static and moving surveillance camera views at up to 1920×1080 resolution and 30 FPS. Designed for violence detection, action recognition, and public safety systems, this surveillance dataset includes rich metadata annotations enabling accurate camera fight analysis and training violence detection models. ## Dataset Structure ### The Numbers Section **Numbers list:** - **Number:** 1,000 — **Text:** Videos ### Tooltip Section **Tooltip items:** - **Name:** Computer Vision - **Name:** Machine Learning - **Name:** Crime - **Name:** Public Safety - **Name:** Action Recognition ### Dataset Info **Table with data:** | Characteristic | Data | | --- | --- | | Description | Videos of simulated physical altercations, captured in controlled environments from static and moving camera perspectives. | | Data types | Video | | Tasks | Public Safety, Computer Vision | | Total number of files | 1,000 | | Labeling | Metadata (fps, resolution, has_audio, environment_type, time_of_day, weather, camera_motion, participants_count, weapon_presence) | **Media Slider:** - **Video on Slider:** **Link to the sample:** [Download sample](https://drive.google.com/drive/folders/1r-QCtXFip286mCTf7k4BhpNkyxQZI-86) ### Technical characteristics **Table with data:** | Characteristic | Data | | --- | --- | | Video Extension | MP4, MOV | | Video Resolutions | 1920 x 1080, 848x478 | | FPS | 30 | **Source and collection methodology:** Source and collection methodology: Data was captured via digital cameras in controlled, simulated environments. ### Dataset Use Cases - слайдер **Industry Cards:** - **Industry:** Public Safety & Smart Cities — **Title:** Real-time violence monitoring in public spaces — **Text:** This fight detection dataset supports video surveillance systems used in streets, transport hubs, and public venues. The dataset contains fighting scenes captured from different cameras, helping detection algorithms recognize violent actions, abnormal activity, and fast motion patterns. It improves automatic violence detection in crowded scenes and real-world surveillance scenarios. - **Industry:** Retail & Commercial Security — **Title:** In-store incident and conflict detection — **Text:** Retail security teams use this surveillance dataset to train models that detect fights, aggressive behavior, and violent interactions inside stores and malls. Labeled video clips with varied camera angles help recognition systems distinguish physical violence from non-violent actions. This reduces false alerts and strengthens proactive loss prevention strategies. - **Industry:** Transportation & Infrastructure — **Title:** Passenger safety and incident response — **Text:** This dataset is well suited for CCTV cameras deployed in stations, airports, and vehicles. It enables detecting violence, group fights, and abnormal events in crowded environments. By learning spatio-temporal features from video footage, detection systems can trigger faster responses to violent scenarios and security threats. - **Industry:** AI Research & Computer Vision — **Title:** Training and benchmarking violence detection models — **Text:** Researchers use this violence detection dataset as training data for action recognition, anomaly detection, and motion analysis tasks. The dataset includes diverse violent sequences and metadata, making it useful for benchmarking detection techniques, testing recognition tasks, and improving model performance across different surveillance systems and camera setups. ### Fact **FAQs Heading:** FAQs **List of Questions:** - **Question:** What should I consider before buying this fight detection dataset? — **Answer:** You should evaluate whether the dataset matches your detection tasks, camera perspectives, and surveillance scenarios. This fight detection dataset is designed for action recognition, violence detection, and anomaly detection using video surveillance footage. - **Question:** What are the sources of data for Unidata fight detection datasets? — **Answer:** The dataset includes only original video footage recorded with digital cameras in controlled, simulated environments and contains no internet-sourced material. - **Question:** What types of annotations are provided with this dataset? — **Answer:** Each video includes structured metadata such as FPS, resolution, audio presence, environment type, time of day, weather, camera motion, participants count, and weapon presence. These labels support detection systems and motion analysis. - **Question:** What video formats and technical specifications are included? — **Answer:** Videos are provided in MP4 and MOV formats at 30 FPS. Resolutions include 1920×1080 and 848×478. - **Question:** Do Unidata datasets comply with GDPR and privacy regulations? — **Answer:** Yes. All datasets are curated in compliance with GDPR and applicable data protection laws. Data is collected from legally permissible sources to ensure ethical and lawful usage. - **Question:** How are Unidata datasets stored and secured? — **Answer:** Datasets are securely stored on AWS cloud infrastructure. Storage and management practices align with ISO 27001 and ISO 27701 standards for security and privacy. - **Question:** Is this a real-world dataset or synthetic data? — **Answer:** This is a real-world dataset with simulated violent events. All videos feature real humans performing staged physical violence, not synthetic or computer-generated scenes. - **Question:** Can this dataset be used to train AI models for public safety applications? — **Answer:** Yes. The dataset provides training data for AI systems designed to identify potentially dangerous situations and support human security teams. It can contribute to automated monitoring solutions that assist operators in reviewing large volumes of surveillance video. - **Question:** How was the fight detection video dataset collected? — **Answer:** The dataset was captured using digital cameras in controlled, simulated environments where physical altercations were recreated for research purposes. This controlled collection approach provides consistent examples for developing and evaluating violence detection algorithms. [Full list of this site's AI-readable pages](https://unidata.pro/llms.txt)