The invention discloses an automatic weighing machine based on deep learning and a program control method thereof. The method relates to an image recognition technology and an internet of things technology based on the depth learning. The method mainly solves the problems of various kinds of artificial weighing articles, price numbers difficult to rememer, low operation efficiency and the like. By constructing the article recognition depth learning model, the weight of the weighing article is collected by the weight sensor and the image of the weighing article is collected. The corresponding information can be inquired out, such as the unit price, the manufacturer, the production date and the like, so as to reduce the workload of the operation personnel and improve the weighing efficiency.