In the context of rapid technological development, the digital transformation of non-heritage culture is significant, which not only helps to protect and inherit, but also attracts more audiences through innovative ways. This study explores non-heritage cultures’ virtual experience interaction design and promotion methods. It investigates the optimization of digital revitalization and interaction design of non-heritage handicrafts with deep learning oversampling techniques to enhance user experience and promote cultural dissemination. After effective screening, 400 valid samples were obtained after collecting 450 questionnaires during the experiment. The interaction design of non-heritage virtual experiences optimized with deep learning technology can significantly improve users’ knowledge and interest in non-heritage culture. The correct rate of participants’ cognition of NRL cultural content increased to more than 95% on average after the experience, in which the proper rate of cognition of plant elements and arrangement forms was as high as 98.9%. The study found that the community interaction function plays a significant role in improving user satisfaction. Deep learning oversampling technology has a substantial effect on the virtual interaction design of non-heritage culture, which can effectively promote the living inheritance and promotion of non-heritage culture.