Cascade algae cell statistical method based on microscope image

Abstract

The invention discloses a cascade algae cell statistical method based on a microscope image, and the method comprises the steps: collecting and marking algae image sample data, and constructing a deep learning model; training the deep learning model to obtain a deep learning detection model; identifying the marked image sample data based on a deep learning detection model to obtain an identification result; and performing cell number statistics on the recognition result based on an image mode recognition technology to obtain a statistical result. According to the method, the deep learning technology and the image mode recognition technology are combined, the deep learning detection model only recognizes the species and coordinates of the algae, cells in the group algae do not need to be concerned, the workload of data labeling is greatly reduced, and the efficiency of model training and model optimization is improved. Meanwhile, the method is wide in application range, the recognition precision of the floating algae is improved, and good expansibility and maintainability are achieved.

Description

technical field [0001] The invention belongs to the technical field of water ecological environment monitoring, in particular to a method for cascading algae cell statistics based on microscope images. Background technique [0002] After using the algae image obtained by microscope and high-definition industrial camera, in order to calculate the algae density, biomass and other related indicators, it is necessary to identify the species of algae and its cell number in the image. Patent Publication Nos. CN109949284A and CN111443028A both propose a method for identifying and counting algae based on a deep learning model, and this method has a better effect on identifying and counting single-celled algae. For group algae, a two-step labeling method is often used, which not only marks the group algae, but also marks the cells inside the group, so that the deep learning model has the ability to identify and count the group algae. This method mainly has the following problems: ...

AI Technical Summary

Problems solved by technology

[0003] (1) The distribution of cells in some colonies of algae is relatively dense. Take Microcystis and Asterella as examples, it is basically difficult to manually label the densely overlapped cells ;

[0004] (2) The workload of labeling cells within the population is relatively large. Whether it is necessary to adjust the labeled samples or add new samples, the corresponding workload is very large

In addition, any adjustment or modification of samples requires retraining of the deep learning model, which leads to poor scalability;

[0005] (3) For some colony algae, there is no way to mark its inner cells, take the genus spp. It is difficult to find a uniform standard to label it accurately;

[0006]⑷When the cells in the group are relatively dense, the deep learning detection model will have more missed detections, which will affect the final statistical results

Images