CN-121982004-A - Tablet antibacterial effect analysis method and system based on deep learning

CN121982004ACN 121982004 ACN121982004 ACN 121982004ACN-121982004-A

Abstract

The invention provides a tablet bacteriostasis effect analysis method and system based on deep learning, comprising the steps of placing tablets on a culture dish for culturing for a preset time, and obtaining an image of the culture dish; and fitting the outline of the drug sensitive ring to obtain the diameter of the outline of the drug sensitive ring. The invention can automatically identify the diameter of the drug sensitive ring in the image, improves the working efficiency of operators and reduces the time of manual comparison.

Inventors

QIN YINGLIN
YANG BIN
LI YANPENG
NIU MIN
HU YIYONG
HUANG DANCHENG
Zhang gun

Assignees

牧原食品股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260128

Claims (10)

1. The tablet bacteriostasis effect analysis method based on deep learning is characterized by comprising the following steps of: after placing the tablets on a culture dish for culturing for a preset time, obtaining an image of the culture dish; Identifying the culture dish image based on a pre-trained deep learning model to obtain a drug sensitive ring profile; Fitting the outline of the drug sensitive ring to obtain the diameter of the outline of the drug sensitive ring.
2. The method of claim 1, wherein the deep learning model comprises a target detection model and an image segmentation model, wherein identifying the culture dish image based on a pre-trained deep learning model to obtain a drug sensitive ring profile comprises: inputting the culture dish image into the target detection model to obtain culture dish information and tablet information in the culture dish image; Inputting the culture dish image into the image segmentation model to obtain the outline of the drug sensitive ring in the culture dish image.
3. The method of claim 2, wherein fitting the drug sensitive coil profile to obtain the diameter of the drug sensitive coil profile comprises: Fitting the drug sensitive ring profile based on the culture dish information and the tablet information, and calculating the diameter of the drug sensitive ring profile based on the fitting result.
4. The method of claim 1, wherein fitting the drug sensitive coil profile to obtain the diameter of the drug sensitive coil profile further comprises: And obtaining a tablet bacteriostasis result based on the diameter of the outline of the drug sensitive ring, and sorting the tablets based on the tablet bacteriostasis result.
5. The method of claim 4, further comprising, after ordering the tablets based on the tablet inhibition results: and drawing the outline of the drug sensitive ring in the culture dish image, and marking the names of the tablets and the ordering of the corresponding antibacterial results of the tablets.
6. The method of claim 2, wherein the object detection model comprises a first convolutional neural network, a region generation network, and a full connectivity layer, and wherein training the object detection model comprises: Acquiring a culture dish sample image, and labeling the culture dish and the tablet in the culture dish sample image through a detection frame based on the culture dish characteristics and the tablet characteristics acquired in advance to obtain labeling data; performing feature extraction on the labeling data through the first convolutional neural network to obtain a feature map; generating a target frame through the area generating network; the target frame and the feature map are subjected to feature fusion through the full-connection layer, so that a culture dish and tablet detection result is obtained; and calculating a first loss value based on the culture dish and tablet detection result and a preset first loss function, and optimizing the target detection model based on the first loss value until the first loss value accords with a first preset condition, so as to obtain a trained target detection model.
7. The method of claim 2, wherein the image segmentation model comprises a second convolutional neural network and mask branches, and wherein the training process of the image segmentation model comprises: acquiring a culture dish sample image, and marking a traditional Chinese medicine sensitive ring area of the culture dish sample image through a contour mask based on the characteristics of the pre-acquired medicine sensitive ring to obtain marked data; Extracting features of the labeling data through the second convolutional neural network, and carrying out feature fusion on the extracted features to obtain a feature map; generating a drug sensitive ring area mask through the mask branches based on the feature map; and calculating a second loss value based on the medicine sensitive loop area mask and a preset second loss function, and optimizing the image segmentation model based on the second loss value until the second loss value meets a second preset condition, so as to obtain a trained image segmentation model.
8. Tablet antibacterial effect analysis system based on deep learning, characterized by comprising: The image acquisition module is used for acquiring an image of the culture dish after the tablet is placed on the culture dish for culturing for a preset time; the medicine sensitive ring identification module is used for identifying the culture dish image based on a pre-trained deep learning model to obtain a medicine sensitive ring outline; And the drug sensitive ring fitting module is used for fitting the drug sensitive ring outline to obtain the diameter of the drug sensitive ring outline.
9. An electronic device comprising a processor and a memory, the memory storing computer executable instructions executable by the processor, the processor executing the computer executable instructions to implement the steps of the method of any one of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, performs the steps of the method of any of the preceding claims 1 to 7.

Description

Tablet antibacterial effect analysis method and system based on deep learning Technical Field The invention relates to the technical field of image processing, in particular to a tablet bacteriostasis effect analysis method and system based on deep learning. Background In the growth process of pigs, the analysis of the drug resistance degree of microorganisms in and out of the bodies of pigs is an important link, and the analysis of the drug resistance degree of microorganisms in and out of the bodies of pigs can not only treat various diseases caused by the microorganisms, but also be used for confirming whether the microorganisms in and out of the bodies of pigs have dynamic drug resistance and the intensity of drug resistance in the growth process, so that the drug type and proportion can be timely adjusted to ensure that the pigs grow healthily at the fastest growth speed. At present, most laboratories only take pictures and archive the microbial drug resistance results, but can not identify and measure the drug sensitive rings in the pictures. Disclosure of Invention In view of the above, the invention aims to provide a tablet antibacterial effect analysis method and a tablet antibacterial effect analysis system based on deep learning, which can automatically identify the diameter of a drug sensitive ring in an image, improve the working efficiency of operators and reduce the time of manual comparison. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: In a first aspect, the invention provides a method for analyzing the antibacterial effect of a tablet based on deep learning, which comprises the steps of placing the tablet on a culture dish for culturing for a preset time, and obtaining an image of the culture dish; and fitting the outline of the drug sensitive ring to obtain the diameter of the outline of the drug sensitive ring. Optionally, the deep learning model comprises a target detection model and an image segmentation model; the method comprises the steps of identifying a culture dish image based on a pre-trained deep learning model to obtain a drug-sensitive ring outline, and inputting the culture dish image into a target detection model to obtain culture dish information and tablet information in the culture dish image, and inputting the culture dish image into an image segmentation model to obtain the drug-sensitive ring outline in the culture dish image. Optionally, fitting the outline of the drug sensitive ring to obtain the diameter of the outline of the drug sensitive ring, wherein the fitting of the outline of the drug sensitive ring based on the information of the culture dish and the information of the tablet and the calculation of the diameter of the outline of the drug sensitive ring based on the fitting result are included. Optionally, after the diameter of the drug sensitive ring outline is obtained by fitting the drug sensitive ring outline, the method further comprises the steps of obtaining a tablet bacteriostasis result based on the diameter of the drug sensitive ring outline and sorting tablets based on the tablet bacteriostasis result. Optionally, after ordering the tablets based on the tablet bacteriostasis results, drawing the outline of the drug sensitive ring in the culture dish image, and labeling the names of the tablets and the ordering of the corresponding tablet bacteriostasis results. The target detection model comprises a first convolutional neural network, a region generation network and a full connection layer, training of the target detection model comprises the steps of obtaining a culture dish sample image, marking a culture dish and a tablet in the culture dish sample image through a detection frame based on the culture dish characteristics and the tablet characteristics which are obtained in advance to obtain marking data, extracting characteristics of the marking data through the first convolutional neural network to obtain a characteristic diagram, generating a target frame through the region generation network, carrying out characteristic fusion on the target frame and the characteristic diagram through the full connection layer to obtain a culture dish and tablet detection result, calculating a first loss value based on the culture dish and the tablet detection result and a preset first loss function, and optimizing the target detection model based on the first loss value until the first loss value meets a first preset condition to obtain the trained target detection model. The image segmentation model comprises a second convolutional neural network and mask branches, wherein the training process of the image segmentation model comprises the steps of obtaining a culture dish sample image, marking a medicine sensitive ring area in the culture dish sample image through a contour mask based on the medicine sensitive ring characteristics obtained in advance to obtain marked data, extracting features of the ma