In view of the problem that the detection cycle of FeO content in sinter is long, which is not conducive to the real-time control of sintering production, a soft measurement model of sinter FeO content based on the image of the empenage section of the sintering and the deep learning algorithm is established, and an online intelligent detection system for sinter FeO content is developed by using the C^# high-level programming language, which has successfully applied to the Zhanjiang 1^# sintering production line of Baosteel. The operation results of the system in the field show that the soft measurement hit rate of the FeO content of the sinter is more than 90% within the error interval of ±0.5%. The system can provide a reference for the real-time control of fuel ratio at the production site, which is of great significance for reducing the fluctuation of FeO content and solid fuel consumption in sinter.