Testing results and comparaison of automated defect recognition software for CCTV inspection

Artificial Intelligence (AI) algorithms applied to underground pipeline inspections bring a new paradigm that promises reduction in the subjectivity associated with coding standard defects and observations. Since the technology and market are maturing and based on the lack of standard certification programs for these technologies, this paper presents a proposed methodology for testing and validation of PACP software programs that use Automatic Defect Recognition (ADR) principles grounded in AI. The testing was conducted to address the specific needs of a municipality but holds relevance for municipalities worldwide. The findings of this testing program enable comparisons with traditional defect/observation coding approaches. The results of the testing and validation program underscore the viability of AI-based inspection methods for multiple purposes, while also delineating their advantages and limitations.