Saccharomyces cerevisiae is one of the most attractive microorganisms, and monitoring changes in its metabolic state at different growth periods is significant for both basic biology and industrial research. In this paper, yeast cells are cultured for different periods based on the yeast generation curve rule, autofluorescence lifetime images are collected using fluorescence lifetime imaging microscopy (FLIM), and an automatic analysis method based on machine learning is proposed, which can rapidly identify young and senile yeast cells without markers. First, a deep-supervised U-Net is applied to automatically segment yeast cells. Then, the features of fluorescence lifetime and morphology of each yeast cell are extracted. Finally, the classification is achieved using the unsupervised clustering method. The experimental results reveal that yeast senescence is accompanied by changes in metabolism. FLIM, as a label-free imaging technique, can be used for the metabolic analysis of yeast cells. When combined with the automated analysis process, it can swiftly and accurately distinguish cells with different metabolic differences, laying the foundation for subsequent screening of single cells.