An artificial intelligence based cell detection method by using hyperspectral data analysis technology includes acquiring a plurality of cells, sampling N images of the cells from a first time point to a second time point, executing an image calibration process for generating N normalized cell hyperspectral images according to the N images of the cells, determining an observation area corresponding to a change of at least one chemical component of each cell during a development period, analyzing the N normalized cell hyperspectral images for generating a key normalized cell hyperspectral image characteristic difference value corresponding to the observation area, and inputting the key normalized cell hyperspectral image characteristic difference value and/or the N normalized cell hyperspectral images to a neural network for detecting qualities of the cells. |