The problem of fluid loss in transport lines through cracks, holes, and clandestine derivations is of interest to various industries, being particularly critical in water distribution systems and hydrocarbon transport. In this context, the detection of fuel theft is of special interest to the industry, as it impacts public safety and the local economy. Therefore, it is necessary to study and develop possible solutions to mitigate this problem. In this context, the proposed methodology aims at detecting localized leaks. For this study, a rotary LIDAR sensor is being used to map the inner surface of a pipeline. The objective of the study is to evaluate some data analysis methods as a tool for processing sensor information and detecting anomalies. Thus, in the present project, it was sought to detect large anomalies through the mapping of the geometry of the duct walls. For this, it was sought to combine the use of LIDAR technology with digital data processing techniques and artificial intelligence. The results obtained were very promising, and large anomalies could be detected with a high level of accuracy. Anomalies of known geometry and dimensions were used as a training set for supervised classifiers, allowing graphical detection from known parameters of the machine learning algorithms used.