AI Technology Streamlines Pipe Inspection

Artificial intelligence provides efficiency and accuracy benefits for inspection programs

AI Technology Streamlines Pipe Inspection

 AI streamlines the pipe inspection process by pairing machine learning algorithms with computer vision to detect and classify sewer defects

Interested in Inspection?

Get Inspection articles, news and videos right in your inbox! Sign up now.

Inspection + Get Alerts

You might be surprised at how little has changed in sewer pipeline inspection over the past 60 years. Closed-circuit television, first introduced in the 1940s, was adopted in the 1950s to determine pipe conditions and line defects. While equipment updates and new standards improved pipe inspections over the years, very few improvements were made to the process of identifying pipeline defects … until now. Artificial intelligence offers a real breakthrough for more accurate, efficient and unbiased inspections.


Utility owners and contractors benefitted from the standardization of pipeline observations by the National Association of Sewer Service Cos. in the early 2000s. The Pipeline Assessment Certification Program meant that pipe segments could be compared within the collection system, regardless of location and who conducted the assessment.

However, standardization did not eliminate variances in observations stemming from subjectivity, bias and experience. For example, an operator with just a few years of experience is likely to capture information differently than an industry veteran; experience levels greatly impact the time the camera is in the pipe and the ease of documentation. Another challenge is how time-intensive the pipeline assessment is both during the inspection and while conducting a quality assurance/quality control review. Operators are expected to manage and review the video recordings of more than 200 NASSCO codes.

The traditional approach typically includes the following four steps:

1. Person-operated camera moves down the sewer

2. Camera captures footage of the sewer’s current state

3. Individual codes defects based on camera footage

4. After results are reviewed, a QA/QC is conducted

These steps often result in the inefficient use of time and dollars. Coding is a very time-consuming step, and because two different people may view a code differently, it is open to interpretation. Enter AI.


Using AI in sewer inspection does not replace the work of field staff. Instead, it supplements their work by providing reliable input that helps them work smarter. Like the human eye, AI depends on quality images to accurately predict what’s in the sewer line, especially when debris and water so easily obscure footage. High-quality CCTV with a high resolution remains essential to providing quality data outputs.

Mimicking the traditional inspection process, AI utilizes existing data and equipment, but with more detailed analysis. Artificial intelligence streamlines the pipe inspection process by pairing machine learning algorithms with computer vision to detect and classify sewer defects that are precursors to backups, overflows and problems that lead to collapses. It captures 80% of PACP codes for structural defects (cracks, fractures, break and holes), O&M (roots and deposits) and construction features (taps).

The artificial intelligence processes both perpendicular and longitudinal views, which provides a more detailed analysis and makes it less likely that defects and other observations will be missed or miscoded. It can also decipher whether the camera is in a manhole or a pipe.

With an AI tool, there is a substantially higher degree of certainty that all incident codes are correctly identified and captured. The AI tool works by utilizing a robust dataset of each computer vision model that is constantly supplemented with inspection data through a supervised learning environment that increases the AI’s accuracy over time. Removing bias from the equation allows the AI tool to reduce errors and boost accuracy rates by identifying if a defect is present and classifying the defect code — all relieving the burden on the operator.


One of the greatest benefits of working with an AI tool is that it gets smarter the more it is used. Each time a contractor or designated QA/QC finds an incorrectly coded defect or defect that has a low probability score, they correct it and retrain the AI. With time, the program gains a deeper understanding of what each defect looks like. The more it’s used, the smarter it becomes, reducing the time required for reviewing sewer inspection data.

Combining AI with human intelligence results in far more reliable and useful datasets, freeing the engineer to make informed decisions and advancing the process of pipeline inspections for the first time in more than half a decade.

Looking ahead, watch for AI efficiencies to cut CCTV video review time in half. The end goal is to enable utility providers to rely on an AI-verified database to make fast, transparent and repeatable data-driven decisions.  


Comments on this site are submitted by users and are not endorsed by nor do they reflect the views or opinions of COLE Publishing, Inc. Comments are moderated before being posted.