The detection of biological spore activity is the basis for effective prevention and control of plant and animal diseases. However, the reduction of its activity level during storage is one of the major problems affecting the application. A rapid and accurate method to detect the activity of biological spores is of great value for exploration and research. In this paper, UV-Vis spectroscopy combined with a one-dimensional convolutional neural network (1D-CNN) is used for the discrimination of dead and viable biological spore. The spectrum of three biological spores were collected and preprocessed by the standard normal variate transformation (SNV).Unsupervised clustering of the sample set was performed using principal component analysis (PCA). The activity discrimination model of biological spores is constructed based on 1D-CNN. The experimental resultsshow that the model has a discriminative accuracy of 100%, which has the potential to replace the traditional methods of determining the dead and viable biological spore.