The conditions in which ship pipelines are located are relatively harsh, and leakage is difficult to avoid. In order to detect small drip faults in time in the early stage of pipeline leakage, and timely maintenance to avoid more serious leakage caused by negligence, a visual monitoring model of pipeline drip leakage is proposed. In this model, the mean background difference method is used to detect pipeline drip, output the characteristic parameters of droplet prospect, combine the virtual coil method to statistically count the number of drip droplets, and finally propose an evaluation scheme for drip volume flow. In order to verify the effectiveness of the model, the method of building a drip experiment bench was used to obtain the drip video to verify the model. The results show that the model can detect pipeline drip droplets well, especially for leaks with slow drip frequency, it can accurately count the number of drips, the flow accuracy is more than 98%, and the relative error of the estimated volume flow is within 20%. This method can effectively monitor pipeline drip leakage and provide reference for the making of pipeline maintenance decisions.