Solutions / Computer vision for your cameras
Computer vision for your cameras
Detection, counting and anomaly detection from camera images, built around your site and your data. We train convolutional neural networks (CNNs) with you, using our own networks or established third-party networks such as YOLO, on your images.
What it is
- Detection and counting: locate and count objects such as vehicles, people or parts.
- Anomaly detection: find deviations from normal in images, for example damage, loose load or missing components on a truck.
- Training on your data: a CNN trained on images from your own site learns your cameras, angles and lighting.
Where it fits
- Sites where off-the-shelf models fail because of camera angle, lighting or weather.
- Inspection tasks where images are already captured but reviewed by hand.
- Operators who need the processing on their own premises.
How a project runs
The approach is scientific: define the measure of success first, then test against it on data the model has not seen.
-
1
Define the task and the cost of errors
What counts as an anomaly, and what a false alarm costs compared with a missed detection.
-
2
Review the data
Coverage of conditions, balance between classes and quality of the labels.
-
3
Train and tune the network
Our own network or a YOLO-family network, trained on your labeled images.
-
4
Evaluate on held-out data
Precision and recall overall and per condition, with the failure cases shown.
-
5
Deploy and monitor
Run it where you need it, and watch performance after rollout.
Confidence threshold explorer
A truck inspection image with eight candidate detections. A detector reports a confidence score for each. Move the threshold to see the trade-off: a high threshold gives fewer false alarms but misses real anomalies; a low threshold finds more but raises false alarms.
Illustrative scene with annotated ground truth. Real thresholds are chosen on your data, according to the cost of each error type.
Where it runs
Our software is used in every case. Pricing is agreed per project after the first meeting.
Your premises, your AI hardware
We install our software on the hardware you already have or buy.
Your premises, our AI hardware
We supply and configure the hardware and our software together.
Our premises
Footage is processed in our lab on our hardware and software, for projects where on-site equipment is not wanted.
Get it your way
NDA available on request.
License it
- Closed source, licensed for your use: device-specific license key or online authentication
- An evaluation version first, so you can test it in your own environment
- License terms agreed to fit your requirements and budget
Have it built
- Designed around your system and your problem
- Built from scratch, or adapted from an existing solution
- Source code and IP stay with you, with documentation and a handover