Computer vision for the cameras you already have.
We turn images into decisions.
Omnicog is a Brazilian artificial intelligence company. We apply AI to existing cameras to detect people, vehicles and events, alert the right team with an image of what happened, and measure how each space is used.
- Since 2021
- Brazilian company, with a team that has more than 10 years in computer vision
- 1,000+ cameras
- in production on highways, through Omnivias
- On site, central or cloud
- runs close to the cameras and keeps working offline
- NVIDIA Inception
- NVIDIA’s program for AI companies
From camera to decision
A chain that is easy to understand and designed to connect to real operations.
Capture
Existing cameras and recorders provide the footage (IP, RTSP, ONVIF).
Analysis
Computer vision models analyze the scenes continuously.
Event
Relevant people, vehicles, objects and situations are detected.
Information
Events become alerts with images, metrics and reports.
Action
The information reaches the right team, VMS, monitoring center or API.
Works with what you already have
Omnicog is an intelligence layer on top of your existing infrastructure: no need to replace your VMS or monitoring center.
IP cameras, DVRs and NVRs
Video over RTSP or ONVIF, standard in most brands.
VMS and monitoring centers
Integration with Digifort and Kria; events with images sent to the monitoring center or the control room.
API and webhooks
Events and indicators for business systems, dashboards and BI.
Mobile and email
Alerts in the app and scheduled PDF reports.
What AI sees in each operation
People, vehicles, objects, areas, flows and events: each project selects the classes and rules that matter to the operation. The camera captures, Omnicog interprets and your team decides.
Smart cities
Occupancy, mobility and urban events for operations centers.
Ports and logistics
People, vehicles, equipment and operational areas.
Airports
Queues, occupancy, trajectories and controlled areas.
Hospitals
Flow and occupancy with privacy by design.
Mining and industry
Hazard zones, machinery, people and vehicles.
Perimeter
Cameras that alert you when someone enters where they shouldn’t.
For highways, our specialization became its own company
Omnivias brings together Omnicog’s experience in traffic analysis (SAT) and automatic incident detection (DAI) for highway concessionaires. It runs on more than 1,000 cameras in production, using the CCTV the highway already has. At the Rota de Santa Maria concessionaire, for example, more than 400 cameras run SAT and DAI, with two years of operation.
- SAT: counting, classification, speed and traffic parameters with no roadwork.
- DAI: stopped vehicles, wrong-way driving, pedestrians, animals on the road, smoke, objects on the road, work zones marked with cones, illegal overtaking and congestion, with images.
Find out what your cameras can do
Short guides for anyone evaluating computer vision in practice.
What is computer vision and how does it work?
From images to classes, events and metrics.
Can you use the cameras you already have?
What to assess before replacing any equipment.
How to reduce irrelevant CCTV alarms
Why context-aware rules beat motion detection.
Edge AI or cloud: where should video be processed?
Latency, bandwidth, cost and continuity.
Assessment of your cameras
The first step is to look at the cameras you already have and tell you what they can do. Then a proof of concept with your own footage shows the result in the field. Depending on the size of the project, both steps can be free of charge, and the proof of concept starts running within 10 days of the agreement between the parties.
Assessment
We review image quality, position, lighting and network of the existing cameras against what you want to detect.
Proof of applicability
We run the analysis on real scenes from your site and measure it against the agreed success criteria.
Deployment
On-premises, central, cloud or hybrid, with rules, alerts, user roles and integrations configured.
Continuous calibration
Rules and models are tuned per camera using the real event history.
Computer vision, made simple
Straight answers to the most common questions from teams evaluating AI for their cameras.
What does Omnicog do?
Omnicog applies artificial intelligence and computer vision to the cameras that companies, industries and cities already have. The system detects people, vehicles, objects, zones and events and turns them into alerts with images, metrics and integrations with your operation.
Do I need to replace my cameras?
Not always. Existing fixed and PTZ cameras can usually be reused when their resolution, angle, lighting and network connection are suitable for the event you want to detect. The technical assessment starts with the cameras already installed.
What is the difference between motion detection and AI analysis?
The recorder’s motion detector reacts to any change in the image: wind in a tree, a shadow, rain, headlights or an animal. AI analysis recognizes what is in the scene (a person, a vehicle, an object) and only alerts when the agreed rule is met, such as “person in the restricted area after hours”. That is why far fewer irrelevant alerts come through.
Do I need to replace the DVR or NVR?
Usually not. Omnicog reads video from IP cameras and recorders (DVR/NVR) over RTSP or ONVIF, standards offered by most brands. During the assessment we check each recorder and each camera.
What can the artificial intelligence detect?
People, vehicles, equipment, objects, occupancy, dwell time, trajectories, queues, line crossings, entry into restricted zones, license plates and events defined for each environment.
Does it replace my VMS or CCTV system?
Not necessarily. Omnicog works as an intelligence layer on top of your existing infrastructure and sends events to your VMS (such as Digifort and Kria), APIs, dashboards, business systems and monitoring centers.
Is the video processed on site or in the cloud?
The architecture is flexible: analysis can run close to the cameras (edge), on central servers, in the cloud or in a hybrid model. When it runs on site, it keeps working even without an internet connection.
Does Omnicog use facial recognition?
Only when the purpose requires it and there is a legal basis and consent. Most applications, such as perimeter, counting and flow, work without identifying anyone, using aggregated data and in line with Brazil’s General Data Protection Law (LGPD).
How many alerts will I receive per day?
It depends on the rules, not on the number of cameras. Each alert corresponds to a rule agreed with you (area, class, time of day, dwell time). During the pilot, rules are calibrated with the real history of each camera so that only what requires action comes through.
What does the proof of concept look like?
It starts with the assessment of your existing cameras. Then we run the analysis on real scenes from your site and measure the result against success criteria agreed beforehand. Deployment only comes when the application shows value. See how the assessment works.
Is the technology developed in Brazil?
Yes. Omnicog is a Brazilian company founded in 2021 that develops its artificial intelligence and computer vision technology in Brazil. The company is a member of the NVIDIA Inception program.
What do you want your cameras to tell you?
Tell us what you need to detect, measure or monitor. We start with the problem, not the technology, and the assessment uses images from the cameras you already have.
- Phone and WhatsApp
- +55 11 3636-8989
- Address
- Al. Terracota, 185, conj. 1213 — São Caetano do Sul, SP, 09531-190