Use Case
Ghegapixel
Ghegapixel is designed to minimize the time needed to create datasets. No more labeling data and converting it to different labeling formats. No more wasting time distributing datasets only to find that a particular error is not properly detected.
Introducing Ghegapixel – the ultimate solution for creating training data with ease.
Our user-friendly interface allows you to generate a diverse range of data with just the touch of a button, thanks to our smart and flexible component building logic. Developed specifically for types of defects in the superstructure, Ghegapixel offers full control and configurability over the data you create. With the ability to generate countless variations, you’ll have plenty of data to kickstart your training and ensure your machine learning models are accurate and reliable.
Railway Specific Parameters (just a few)
- Gradation curve or track ballast (Schotter Sieblinie, Körnung)
- Track curve Radius
- Sinusoidal cart motion
- Cart length, bogie placement, etc.
- Polished rail head
General Parameters (just a few)
- Dirt
- Trash
- Rust
- Moisture
- Defects (Missing, Misaligned, Fractured components etc.)
This system allows us to achieve unprecedented levels of accuracy in detection models.
Image
Panoptic Segmentation
Depth Map
Endless Variations
