Avitas Systems provides advanced analytics that incorporate computer vision, machine learning, and physics-based modeling.

3D Modeling

3D asset representation for accurate data mapping and context for analytics:

  • Photo-realistic models automatically generated from RGB and IR images
  • Point cloud models from LiDAR scans

View Info Sheet

3D Understanding

Automated analysis of 3D data to extract findings, including encroaching vegetation

Deep Learning Image/Video Understanding

Using deep learning techniques such as CNNs, LSTMs, GANs for classification of objects and defects in multi-modal images and videos, as well as:

  • Abnormal heat transfer rates in IR video to detect indicators of corrosion under insulation
  • Change in appearance of an asset/component over time as observed in recurring inspection images
  • Detection of gas plumes and sizes, oil leaks and sizes, heat spikes

4D Asset Monitoring

Time series data extraction from recurring external visual inspections of the same asset

  • Use computer vision techniques to associate image and video findings back to the asset 3D model
  • Decompose 3D asset models using 3D segmentation techniques as needed to match the asset hierarchy for accurate cataloging findings over time

Predictive Modeling Using Hybrid Techniques

Predicting outcomes using advanced data-driven models, combining machine learning with subject matter expertise for physics-based feature engineering:

  • Predicting corrosion under insulation in pipelines and plants
  • Predicting pitting corrosion in pipelines
  • Predicting unplanned shutdowns for plant operations


Learn more about our bespoke cloud-based platform that supports the full end-to-end inspection management process