Aiming at the problem of low diagnostic accuracy of the intelligent diagnosis method caused by many application scenarios of rolling bearings in packaging machinery and the difficulty of collecting effective fault data, a data-enhanced intelligent diagnosis method of rolling bearings is proposed. First, according to the fault characteristics of bearing vibration signals, a data enhancement method was proposed to effectively expand the diversity of training data samples. Then, the convolutional neural network was used to train the original samples and enhanced samples for fault diagnosis, so as to greatly improve the diagnosis performance of the diagnosis model. In order to verify the effectiveness of the proposed method, a rolling bearing failure test rig was established and bearing failure data were collected. The experimental results show that when the label training samples are insufficient, the proposed method has a greater improvement in diagnostic accuracy than the method without data enhancement, and can accurately identify various bearing faults. This method realizes the accurate fault diagnosis of the rolling bearing under the scarce marked samples, and provides a reliable method for ensuring the diagnosis accuracy of the rolling bearing fault diagnosis of the packaging machinery.