The invention belongs to the technical field of biomedicine, and discloses a deep learning-based chromosome uploidy prediction system and method, terminal and medium, and the method comprises the steps: selecting a part of embryo images in the whole process of embryo development, combining the images into a video as a training data set according to the time sequence, building a video sequence feature network model, training the embryo video, extracting morphological characteristics and dynamic characteristics in the embryo development process; and fusing the embryo video features and the clinical data features of the patient, and training a network according to the fused features to obtain a prediction result. According to the method, a video sequence feature network framework is creatively used, the space-time characteristics in the embryo development process are well obtained, and the extraction of the dynamic characteristics and the morphological characteristics of the embryo cells is effectively completed; the prediction model provided by the invention can automatically complete the prediction of the chromosome uploidy, and no manual intervention exists in the prediction process.