Synthetic Computer Vision Datasets
Ready-to-train computer vision training data for object detection, security AI, drone and aerial vision, and safety-critical models. Every dataset is 100% synthetic, YOLO-annotated, and GDPR-safe — with a free open-source sample on Kaggle and Hugging Face.
Browse synthetic computer vision datasets
UAV & Aerial View: Battle Tank Detection Dataset
Synthetic aerial and drone-perspective dataset for detecting main battle tanks and heavy armored vehicles. YOLO-annotated, 100% synthetic, ready for defense and reconnaissance AI training.
Full Package
- 3,000 images
- YOLO annotations
- Image dataset
CCTV Shoplifting Detection Dataset (YOLO + VLM)
Synthetic CCTV dataset for shoplifting and concealment detection in retail. Person detection with pose keypoints, plus VLM-ready captions and video sequences for temporal action recognition.
Full Package
- 5,000+ frames + 100+ videos
- YOLO annotations
- 8 eval videos
CCTV ATM Robbery Detection Dataset: Gun & Knife
Synthetic surveillance dataset for detecting armed robberies at ATMs. Multi-class YOLO annotations distinguish offender from victim and gun from knife, captured under realistic high-angle CCTV conditions.
Full Package
- 3,000 images
- YOLO annotations
- Image dataset
CCTV Smoke & Fire Emergency Detection Dataset
Early-stage fire detection dataset featuring small ignition points, bin fires, and smoldering debris from a surveillance perspective. Designed to train models that catch fires in their first 60 seconds — when intervention matters most.
Full Package
- 2,500 images
- YOLO annotations
- Image dataset
Long-Distance Wildfire & Smoke Detection Dataset
Synthetic long-range wildfire and smoke detection dataset captured from elevated vantage points — ridgelines, lookout towers, and fire-watch cameras. Designed to spot incipient smoke plumes against complex terrain at distances of several kilometers, where every minute matters.
Full Package
- 1,500 images
- YOLO annotations
- Image dataset
CCTV Weapon Detection Dataset
Synthetic training data for AI concealed weapons detection from CCTV camera angles. YOLO-annotated images with person and weapon classes, covering handguns and rifles in overhead, wall-mounted, and doorway camera perspectives.
Full Package
- 1,000+ images
- YOLO annotations
- Evaluation videos included
CCTV Knife Detection Dataset
Specialized synthetic dataset for detecting people holding knives in CCTV security camera perspectives. Includes diverse scenarios with precise YOLO annotations for person and knife classes, optimized for security and surveillance applications.
Full Package
- 1,200 images
- YOLO annotations
- Evaluation videos included
CCTV Fall & Incident Detection Dataset
Synthetic dataset for fall detection and incident monitoring from overhead CCTV perspectives. Features dual annotations with bounding boxes and 17-keypoint skeletons (COCO standard), enabling models to understand human posture and distinguish between standing and fallen individuals. Compatible with YOLOv8-Pose and YOLO11-Pose models.
Full Package
- 1,200 images
- YOLO annotations
- Evaluation videos included
Airport Drone Threat & Safety Dataset
Synthetic dataset for aviation security and rogue drone detection. Detects small drones against complex airport backgrounds—terminals, hangars, tarmac—where real training data is nearly impossible to capture due to flight regulations. Features adverse weather conditions (fog, rain, night), small object focus (<5% of frame), and 100% GDPR compliant synthetic imagery.
Full Package
- 1,200 images
- YOLO annotations
- Evaluation videos included
CCTV Aggressive Pose Detection Dataset
Specialized synthetic dataset for detecting aggressive body poses and threatening behaviors in CCTV footage. Includes keypoint annotations for precise pose estimation, ideal for security systems and threat detection applications.
Full Package
- 1,000 images
- Keypoint annotations
- Evaluation videos included
Pedestrian Attribute Recognition Dataset (Pedestrian-1K)
Synthetic CCTV dataset for Pedestrian Attribute Recognition (PAR) and Re-Identification (Re-ID). Features natural language descriptions and structured attribute maps for Vision Transformers (ViT) and CLIP-based models. High-angle surveillance perspective with realistic noise and compression artifacts. 100% GDPR and CCPA compliant.
Full Package
- 50,000 images
- YOLO annotations
- Image dataset
CCTV Weapon Detection: Rifles vs Umbrellas
Synthetic dataset designed to solve the biggest problem in weapon detection AI: false positives. Features a balanced split of rifles (threat) vs umbrellas (confuser) from realistic CCTV angles. Includes hard negative mining focus, sensor noise, motion blur, and varied lighting. YOLO ready and 100% privacy-compliant.
Full Package
- 1,000 images
- YOLO annotations
- 8 eval videos
Weapon Detection: Handgun vs Bag of Chips
Synthetic dataset designed to fix the 'Doritos Effect' — AI false positives on snacks and shiny objects. After a student was handcuffed because an AI flagged a bag of Doritos as a handgun, this dataset uses Hard Negative Mining with a balanced split of handguns (threat) vs chip bags (confuser) from realistic CCTV angles. Includes sensor noise, motion blur, and varied lighting. YOLO ready and 100% privacy-compliant.
Full Package
- 1,000+ images
- YOLO annotations
- Evaluation videos included
Military Drone Swarm & Saturation Attack Dataset
Synthetic dataset designed for Counter-UAS (C-UAS) saturation attacks. Features high-density drone swarms (5–15 units per frame) in complex, photorealistic environments. Engineered to benchmark Small Object Detection and Swarm Tracking algorithms where standard models fail.
Full Package
- 1,000+ images
- YOLO annotations
- 5 eval videos
Drone Civilian / Armed Military Detection Dataset
Advanced synthetic dataset for detecting and distinguishing between civilians and armed military personnel from drone/UAV perspectives. Features diverse terrain and scenarios with YOLO annotations optimized for aerial surveillance and defense applications.
Full Package
- 800 images
- YOLO annotations
- Evaluation videos included
View Gallery (1)
Driver Drowsiness Detection Dataset
Advanced synthetic dataset for detecting driver fatigue and distraction through precise eye monitoring. Building AI that detects if a driver is tired or distracted is difficult with real-world data, where labeling eye openness is often inaccurate. This synthetic dataset solves the accuracy problem by providing computer-generated images with exact eye positions and states down to the pixel, featuring eye keypoint annotations and 20 validation videos.
Full Package
- 10,000 images
- YOLO annotations
- Evaluation videos included
Need a Custom Dataset?
We can create custom synthetic datasets tailored to your specific use case and requirements. Contact us to discuss your needs.