Traffic light detection: Datasets & Models

For teams building autonomous driving stacks, ADAS features, or smart city traffic management, manually reviewing footage to catalog signal states is slow, inconsistent, and impossible to scale to live operation. Automating traffic light detection with computer vision gives you reliable, frame-by-frame recognition that feeds directly into navigation logic, violation monitoring, or signal timing analytics. It removes the human bottleneck and delivers repeatable results across thousands of intersections and hours of video.

You get consistent detection accuracy across day, night, and adverse weather, and the models integrate with existing traffic cameras, dashcams, and edge devices without new hardware. Because the workflow is built on a standard detection pipeline, it scales from a single test vehicle to a citywide deployment, and RF-DETR provides strong accuracy on small and distant signal heads common in real road scenes.

1. Upload an image: Drop any photo into the widget above to try traffic light detection for free, right on this page.
2. Preview instant results: The model detects red light, green light, yellow light, traffic light, pedestrian signal 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.
3. 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 on your own images for production accuracy.
4. 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.