The invention discloses an AI intelligent multi-terminal adaptive non-invasive
hemoglobin detector and a detection method, and the
detector comprises a configuration module which is used for transmitting combined light with a specific
wavelength to penetrate a fingernail
bed of a user when the fingertip of the user is placed in an
accommodation cavity; the
image acquisition module is integrated in the intelligent terminal and is used for shooting a multiband image reflected by a fingernail
bed when the fingertip of the user is placed in the accommodating cavity; the transmission module is used for transmitting the multiband image to the
processing module; and the
processing module is internally provided with a
deep learning spectral feature regression model and is used for extracting features related to the
hemoglobin concentration from the multiband image, carrying out regression calculation and outputting a quantitative
hemoglobin concentration value. Optical signals carry complete concentration related information after penetrating through the fingernail
bed, so that the detection precision is improved;
image acquisition is integrated on an intelligent terminal commonly used by a user, the use threshold is reduced through multi-terminal
adaptation, the accuracy of data
feature extraction is improved based on a
deep learning spectral feature regression model, and then the accuracy of a detection result is improved.