Crowd density estimation: Datasets & Models
A computer vision workflow that segments crowds in images and video to estimate density levels across public spaces. Built for security and surveillance teams monitoring foot traffic, gatherings, and occupancy.
Crowd density estimation
This crowd density detection workflow applies instance segmentation to outline individual people in images and video, then converts those segments into per-zone counts and density estimates across the monitored area. It accepts standard camera feeds or still images as input and returns segmented crowd regions along with density levels that indicate where occupancy is rising toward defined limits.
Try it out
Powered by general detection model
Run on custom image
Drop an image here or click to upload
Why automate crowd density estimation
Manually watching a wall of monitors to judge when a plaza, platform, or stadium concourse is getting dangerously packed is slow, subjective, and easy to get wrong during the exact moments it matters most. Automating crowd density detection with segmentation gives security teams consistent, quantitative readings on occupancy and movement, and can trigger alerts before a crush or bottleneck develops. It removes the guesswork from staffing decisions, gate control, and emergency response. For surveillance operations covering many cameras at once, one model can watch every zone continuously without fatigue.
Key benefits
Instance segmentation delivers accurate counts even in dense, overlapping crowds where box-based detectors break down, translating directly into reliable density thresholds and earlier alerts. The workflow runs on your existing IP and CCTV camera infrastructure, so there is no new hardware to install, and it scales from a single entrance to a facility-wide network of feeds. Density maps and per-zone metrics integrate into dashboards and alerting systems your security team already uses.
How to use it
- Upload an image: Drop any photo into the widget above to try crowd density estimation for free, right on this page.
- Preview instant results: The model detects person, head, crowd, backpack, umbrella within seconds and returns an annotated image with structured predictions, so you can immediately see how well it works. Adjust the detection classes and re-run until the results fit your scenario.
- Like what you see?: With a free Roboflow account, you can deploy this workflow with the Serverless Cloud API, customize it in your workspace, or train RF-DETR (instance segmentation) on your own images for production accuracy.
- Get started for free: No code or training data required to begin: upload, review the results, and make it part of your stack in a single click.
Recommended model for crowd density estimation
RF-DETR is state of the art for real-time object detection, instance segmentation, and keypoint detection, with higher accuracy and lower latency than comparable models, and it's Apache 2.0 licensed. When you're ready to build a production model for crowd density estimation, fine-tune it on your own labeled images, then deploy to a Serverless Cloud API endpoint.
Not sure where to start?
Describe your Crowd Density Estimation project to Roboflow Agent.