FAQ

PipeAId is a Computer Vision Machine Learning AI system that is able to identify defects and observations that occur in a sewer pipe.

If you think of PipeAId in human terms, the machine learning is the brains and the computer vision is the eyes. As more inspection video is processed, PipeAId gets smarter within a supervised learning environment.

PipeAId is designed to fit in with equipment and operating procedures that are typical in the sewer inspection industry. This means that owners and contractors do not have to re-train their staff on how to inspect pipes, which eliminates any learning curve for field staff.

In many cases, there is a fine line between types of observations such as cracks/fractures and fractures/breaks. Incorrect coding could mean inadvertently noting that a section of pipe needs repair work when the condition does not warrant it or noting that a pipe does not need repair when it does. PipeAId removes human bias and subjectivity to ensure that incidents are coded correctly and the pipe’s condition is accurate.

PipeAId is trained on millions of lineal feet of pipe. With this information the AI is capable of identifying the presence of pipe defects 100% of the time. Once a defect is detected, PipeAId assigns a NASSCO code for 90% of the most common observations and uses NASSCO PACP trained technicians for any remaining.

Utilizing AI in our workflow generates more reliable results that are repeatable over time and can aid in determining how quickly your assets are deteriorating. It also allows you to make data-driven decisions on repair, replacement or maintenance activities. Reliable results mean that you can quickly identify bad pipes in your system with 100% accuracy and prioritize repairs.

PipeAId is able to process any inspection video, coded or not, as long as it includes length of pipe. It is not dependent on any specific inspection software.

PipeAId does not require an existing database. PipeAId can create a database of a pipe’s metadata such as size, material, upstream/downstream manholes, date and direction of survey, etc. via a text recognition component that captures data typically displayed at the start of an inspection video.

PipeAId has moved one step past the typical PDF reports generated by sewer inspection software by incorporating your GIS into our deliverable. This allows us to show both the condition of your sewer assets and the geospatial location of each defect along the pipe.