Closed circuit television (CCTV) is probably one of the most important technologies that is used by municipalities in order to monitor the structural and operational condition of sewer pipes. To be useful, CCTV video footage needs to be collected according to standards, which make such an operation, time consuming especially when pipes have operational issues like debris or tree roots. In this respect, developing benchmarks for data collection can be an important source of information that can improve the efficiency of future surveying campaigns. Computer simulation is an effective method for improving the efficiency of maintenance work schedules. However, CCTV collection data consists of abundant noise (waiting time or defect inspection time) due to the characteristics of pipes in different structural or operational conditions. For example, crawlers equipped with CCTV cameras could be blocked by deposits or serious structural issues in the pipe, which would cost some waiting time for the crawler to proceed with the inspection. In order to extract the standard CCTV collection time, excluding waiting time and defect inspection time a machine learning based approach is proposed in this work in the form of an algorithm commonly known as the Random Sample Consensus (RANSAC). This algorithm is developed to clean the data automatically, arriving at a function of CCTV collection time with two variables (i.e., length of pipe segment and number of taps in the pipe). The results can be fed into a simulation model to imitate the CCTV collection work in future research.