For engineering teams in smart transportation or road mapping, building a road sign recognition model used to feel like climbing a technical mountain. Coding skills, time-consuming data labelling, and debugging—enough to make even experienced teams hesitate.

But here’s the game-changer: our tool skips all that. Just mark road sign templates once on point clouds—outline edges and key points—and you’re set. No coding, no repeated training. The model’s already prepped, ready to go.

How does it work? Simpler than you think

Traditional self-training? A maze. You’d need Python and TensorFlow skills to even start. Labeling? Guesswork. Hours of marking random points, with no guarantee that the model would learn.

Our tool cuts through that.

Take road signs—turn arrows, straight-ahead markers. You find them in your point cloud data, mark their edges and a few key points once. That’s it. The pre-trained model instantly recognises the patterns, with no extra steps.

Maximize image

Edit image

Delete image

Real results from Australia

The School of Civil Engineering at the University of New South Wales in Australia tested this function and achieved promising results, with their project data collected by GreenValley International's Mobile Mapping System (MMS) LiMobileM2 and processed by software LiDAR360MLS.

Minimize image

Edit image

Delete image

Minimize image

Edit image

Delete image

Minimize image

Edit image

Delete image

Minimize image

Edit image

Delete image

They labelled turn arrows, straight markers, and other signs once, just outlining edges and key points.

Minimize image

Edit image

Delete image

Minimize image

Edit image

Delete image

After marking all the road markings on this route—with each road marking marked only once—start the automatic training using the software's built-in model. The training accuracy for object detection is 99%, while the training accuracy for keypoint detection is 91%.

Minimize image

Edit image

Delete image

After automatic training, how can we verify the model’s accuracy? The software LiDAR360MLS can generate an accuracy report for this purpose.

Minimize image

Edit image

Delete image

Then, it can automatically recognise the markings along the entire road. Here is the result.

Minimize image

Edit image

Delete image

Minimize image

Edit image

Delete image

Minimize image

Edit image

Delete image

No prior coding experience, no tedious training loops. The model kicked in immediately, with accuracy that met their project needs.

Reviews from Australian clients: “We used to spend weeks prepping a usable model. Now? It’s done in hours. Game-changer for field mapping.”

For teams tired of technical hurdles

If you’re also frustrated by the hassle of annotation, you really should give this a try.

Anyone in tech knows that avoiding detours is the ultimate efficiency. This tool isn’t complicated to use — it simply helps make road markings "visible" in point clouds and lets the model learn them thoroughly in one go.

If you’re constantly held back by repeated annotations slowing down your progress, why not give it a shot? After all, who wouldn’t want to finish annotating data early and get off work on time?

Click here LiDAR360MLS Point Cloud Feature Intelligent Extraction and Analysis Software  to apply for a trial and experience the power of LiDAR360MLS firsthand! 

Interested in the high-precision MMS LiMobile M2?

Book a demo today at info@greenvalleyintl.com