Zero Sample Image Retrieval Technology This paper proposes a digital media technology for zero sample image retrieval based on the teaching methodology of modern art education classrooms. Then, it studies the zero-sample semantic segmentation network based on spatial embedding and researches the hidden space selection and construction, as well as related loss functions and parameter balance. As a demonstration, an empirical study is conducted in a modern art education classroom at a school. According to the experimental results, 45.3% and 49.6% of students and teachers acknowledge the use of zero-sample image retrieval technology in modern art education. In the quantitative study of the impact on students’ aesthetic ability, there is no significant difference between the scores of students’ aesthetic ability pre-test dimensions in the experimental and control groups in all five dimensions (p>0.05) and the dimension of aesthetic perception ability, the difference between the scores of the experimental and control groups in the pre-test is only 0.01±0.12.