Bowl plugs were mainly used for plugging auxiliary holes on the engines, the processing precisions were high,required diameter tolerance was micrometer. At present the bowl plugs mechanical errors were mainly manual detections, manual detections precisions were low and can not meet the requirements of modern production online detections. Therefore proposes a method of used binocular vision to detect bowl plugs machining errors, put the bowl plugs on a rotating mechanism for constant speed rotations, two parallel cameras acquire multi groups diameter end face images. Used the proposed end face impurity removal method calibration test results, through data processing to obtain bowl plugs diameters and tapers mechanical errors. Set up a binocular vision online detection system for bowl plugs machining errors. The test results show that this method can meet the bowl plugs detection requirements and achieve online detections.