Ultrasonic sensors are being used to detect the presence of wall thinning in oil and gas pipelines. The ultrasonic waves travel through the pipe material and the presence of defects will cause varying changes to the waves. These changes will be used to detect the type and location of the defects. The high frequency of the ultrasonic signals and the presence of noise make it necessary for high quality data acquisition systems and efficient processing techniques. This paper investigates pipeline defect detection using a data acquisition (DAQ) systems constructed using an off-the-shelf A/D converter. The data from these sensors will be classified using a powerful machine learning algorithm called Support Vector Machines (SVM). A machine learning tool is needed to here to identify the minor characteristic of ultrasound signal that changes with the presence of defects and thus removing human decision making as a factor. Results show that that the self-fabricated DAQ and Support Vector machine can detect the presence of wall thinning with a suitable degree of accuracy.