| CPC G06V 20/695 (2022.01) [G06V 10/774 (2022.01); G06V 10/776 (2022.01); G06V 20/698 (2022.01)] | 10 Claims |
|
1. A cell selection device, comprising:
a pre-processing unit adjusting a size of a cell image extracted from image data obtained by capturing a cell into a preset size, converting the resized cell image into an HSV color space (hue, saturation, value model) to correct a brightness value, and normalizing a pixel value to generate training data;
a model generation unit generating a second learning model by performing transfer learning using at least two different pre-trained first learning models and combining outputs of the first learning models based on a similarity between the training data and the datasets of the first learning models;
a cell classification unit determining a homogeneity of the cell based on the number of specific single cells inferred as functional cells using the second learning model from the cell image and classifying the cell into a positive cell and a negative cell based on the homogeneity; and
a visualization unit marking a portion serving as a classification reference on the cell image based on a result of the classification by the second learning model and outputting the marked image to indicate a basis of the classification.
|