RailLens
by
Drag and drop images here or click to select files
For best results, ensure your image matches our training data. The rail should be centered with only one side visible, and the image should include one sleeper. Avoid images that are too close, too far, or show both sides of the track.
For best results, ensure your image matches our training data. The rail should be centered with only one side visible, and the image should include one sleeper. Avoid images that are too close, too far, or show both sides of the track.
At ex-nihilo, we are dedicated to developing the world's most advanced defect detection software, even for rare anomalies like fractured and misaligned tension clamps.
RailLens showcases the first model capable of detecting these defects on images of ballastless track with unparalleled accuracy (F1 Score of 0.95).
This is made possible by our proprietary Ghegapixel technology, which generates high-quality synthetic data to train our AI models effectively.
Learn more about our technology at ex-nihilo.ai .