A digital unmanned farming system for pig was developed based on the integration of machine vision， big data， the internet of things （IoT）， intelligent modeling， and database technologies to solve the problems including the low efficiency of production， manual removal of manure， large size of equipment， high consumption of power， and the low accuracy of perception data faced by the majority of pig farms in China. A demonstration base for applying digital unmanned systems in pig farming was established. The digital unmanned technologies including intelligent visual monitoring of pig farming， remote operating system terminals， digitalized remote environmental monitoring IoT perception and intelligent scientific decision-making， automatic operation reminders， and the application of IoT intelligent devices were completed. A comparative experiment between the artificial farming and the digital unmanned farming was conducted by establishing application scenarios. The results showed that the established system realized non-contact pig face recognition through RKNet combined with HAM mechanism， with the accuracy of the pig face recognition model of 99.26%， the precision of 99.20%， and the model size of only 1.52 MB. It was deployed in embedded systems. The application of digital unmanned systems in pig farming reduced the average daily working hours of farmer from 4 h to 2.5 h， increased the average daily growth weight of pigs from 1.21 kg to 1.72 kg， and reduced the average daily abnormal behavior of pigs by 36.4%. It is indicated that establishing a digital unmanned system for pig farming can improve the efficiency of farming pig， reducing costs of labor， and increasing the economic benefits of the pig farming industry.