Shorter queues from the cameras you already have
Count, track, and alert in real time with Ultralytics YOLO, trained on your own footage in Ultralytics Platform and licensed for your product.
Trusted by the world's leading organizations
Built for teams that ship queue analytics
Ultralytics provides the models, tools, and license. You build and own the product, from alerts to dashboards.
Retail and footfall analytics vendors
Move from LiDAR and 3D counters to the CCTV your customers already have, and add queue alerts, dwell time, and occupancy to your product.
System integrators
Deliver queue and occupancy projects for retailers, banks, airports, and venues on mixed camera fleets, fully on-premises.
Camera, VMS, and in-house teams
Embed YOLO in your devices and video management software, or build queue analytics for your own chain of stores.
Cali Intelligence shortens checkout queues by 43% with Ultralytics YOLO
See how Cali Intelligence uses Ultralytics YOLO on existing store CCTV to cut average checkout queue length by 43% with real-time alerts and forecasting.

How it works
From the cameras you have to an alert your staff can act on.
Connect existing cameras
Use RTSP streams or VMS feeds from cameras already installed, including mixed brands and ceiling or fisheye views.
Count and track at the edge
YOLO detects and tracks every person inside each queue zone on a Jetson, Hailo, or Intel box at the site.
Alert before the queue builds
Tell staff to open another lane when a zone passes its threshold, and log queue length and wait time for reports.
Fine-tune and roll out
Train on footage from a few representative sites in Ultralytics Platform, then deploy one model to every location.
Everything a queue camera can measure
Five YOLO tasks in one model family, trained on your footage and deployed the same way.
Try YOLO26 inference
Drag and drop an image to see real-time object detection
Why teams build on Ultralytics YOLO
Real-time counting on your own hardware at a fixed cost, with no per-call API fees.
Runs at the edge
Export to TensorRT, OpenVINO, ONNX, and more, then run on NVIDIA Jetson, Hailo, Intel CPUs, or a branch server.
Private by design
Count people without identifying them. Video stays on site, and only counts and events leave.
One model for every site
Fine-tune once on representative stores, then roll the same model out to every location and camera angle.
Two ways to start
Train and deploy on Ultralytics Platform, or add YOLO to your own pipeline.
Ultralytics Platform
Upload footage from your sites, label it with AI-assisted annotation, train a custom YOLO model, and export it for your edge hardware, all in one place.

Python
The QueueManager in the Ultralytics package counts tracked people inside your zone in a few lines. License YOLO to ship it in your own product.
from ultralytics import solutions
# Outline the waiting area once per lane
queue = solutions.QueueManager(
model="yolo26n.pt",
region=[(20, 400), (1080, 400), (1080, 360), (20, 360)],
)
results = queue(frame) # a frame from your camera or VMS
print(results.queue_count) # people waiting nowEnterprise-grade security and compliance
Build and deploy with confidence. ISO 27001 certified, SOC 2 attested, with support for HIPAA compliance.
Frequently asked questions
Yes. YOLO works on standard RTSP streams and VMS feeds from mixed camera brands, including ceiling-mounted and fisheye views. For unusual angles or lighting, fine-tune on your own footage in Ultralytics Platform.
You build and own the application. Ultralytics provides the YOLO models, the QueueManager and other counting solutions, Ultralytics Platform to train on your footage, and the commercial license. Your team builds the alerts, dashboards, and integrations on top.
YOLO runs on NVIDIA Jetson, Hailo and Axelera accelerators, Intel CPUs with OpenVINO, and GPU servers. Choose the model size by stream count: Cali Intelligence processes 3 to 6 camera streams at around 3 FPS each on compact on-site servers.
Pretrained YOLO26 detects people well out of the box. Segmentation and tracking keep counts stable when people overlap, and fine-tuning on your own stores handles occlusion, carts, and lighting. In production, Cali Intelligence keeps a miss rate below 6% across different store layouts.
Both. The zone count gives queue length, and tracking measures how long each person stays in line. Length matters most: a Management Science study of supermarket deli queues found that shoppers react to the line they can see, and that queues of 15 or more people cut sales by about 10%.
Yes. Inference runs on your own hardware, so video stays on site and only counts or events leave. Queue counting detects people without identifying them: no facial recognition and no matching people across visits.
Yes. Train a custom class for staff uniforms or work areas in Ultralytics Platform, then exclude those detections from queue counts.
Ultralytics YOLO is open source under AGPL-3.0, so you can prototype freely. To ship it in a closed-source product or service, an Enterprise License removes the AGPL obligations. Enter your work email above to get a quote.
More use cases

Retail loss prevention

Inventory management

Visual inspection

Driver monitoring

Document processing

Defect detection
Your vision, realized
Join millions of developers building scalable solutions for smart use cases across industries. Annotate, train, and deploy — all from Ultralytics platform.



