AlgaeAI 700 plankton automatic classification and counting system

Quick numbersAlgaeAI 700The automatic classification and counting system for planktonic organisms, newly developed by XunshuAlgaeAIThe system consists of a fully automated scanning microscopy system. With the help of a high-speed three-axis electric platform, obtain high-definition microscopic images and utilize a super deep learning based approachAlgaeAITechnology can automatically identify species, classify and count, measure biomass, calculate cell density, generate and export data reports for plankton in water bodies, and achieve electronic recording of plankton to ensure the integrity of electronic data.AlgaeAI 700It is also equipped with an intelligent identification expert module for plankton, providing an important tool for the study of plankton diversity in aquatic ecosystems.

Artificial intelligence, within reach

AlgaeAI 700It is an advanced intelligent analysis and recognition computing system for plankton using Xunshu's ultra deep machine learning. Personnel conducting plankton analysis can use excellent data analysis techniques No deep expertise in planktonic biology is required.

l Simple, one click completion of identification, classification, counting, percentage, algae density calculation, dominant algae sorting, and report output

l Fast, single field analysis only requires0.6second

l Accurate and recognizable, trained by neural networks145Common algae,60Species of freshwater plankton with high recognition accuracy95%above

l High robustness, suitable for common complex microscopic images in experiments: overlapping algal cells at high densities, local structural blurring caused by insufficient depth of field, background mixed with many impurities, and cells only partially within the field of view…….

Full transparency, visible reality

The analysis, recognition, and statistical process are fully displayed on the screen, and the operator can clearly observe the analysis process and processing results of each image.

l click“AIstart”The main window images flashed one by one, and the algae cells were framed one by one, with the name of the algae on the top of the frame

l The green scrollbar on the upper right indicates that the sample is currently being tested……

l On the right is real-time jumping updated data: phylum, algae name, algae quantity, percentage, algae density……

Detection completed, click on the image queue in the bottom left corner to easily view the algae count on each image: individual names are incorrect? Individual algae not detected? Do individual algae need to be removed? Simple, click on the toolbox,2-3It can be corrected in seconds.

Solving the difficulties of detecting complex field of view images

There are many challenges in the quantitative analysis process of phytoplankton, such as high suspended impurities in the collected quantitative water samples, high cell density after concentration, overlapping microalgae, existence of layered and divided fields of view of algal cells in the counting box, insufficient clarity of microscope optical imaging, and inaccurate focusing……

AlgaeAI 700By utilizing its super deep machine learning capabilities, high-quality analysis and recognition results can be generated quickly and reliably.

Not afraid of the crossing and overlapping of algae cells, it can be automatically untied directly

Algae density exceeds10¹⁰cells/LSamples with cluttered backgrounds can still be recognized and counted

The light source and focus adjustment are not in place, and brittle rod algae and disk star algae are very shallow and light, but can still be recognized

Automatic identification of planktonic animals

The digital treasure trove of intelligent identification of algae

The grand algal image library covers freshwater and marine species in inland waters such as rivers, lakes, reservoirs and surrounding waters in China. The exquisite selected pictures and text introduction, combined with a rich retrieval framework, are helpers for the basic teaching of plankton and the popularization of algal knowledge by water environment monitoring organizations.

Taxonomic search, key editing and column introduction of common algae, with clear understanding of morphology, structure, reproduction, and ecology.

Morphological retrieval, based on morphological similarity and gradient, combines and classifies into graphic language, and combines the structural characteristics of cells or populations, such as flagella, pigment bodies, patterns, gelatinous covers, etc., to achieve accurate and fast morphological retrieval.

High quality microscopic scanning imaging

Olympus research grade biological microscopeBX43For optical imaging carriers, configure smooth and quietXYZElectric stage, achieving one click precise operation: automatic focusing, automatic scanning, excellent image quality.

Main functions and technical indicators

1. Analysis standards

Compliant with《SL733-2016Technical Regulations for Monitoring Phytoplankton in Inland Waters and Technical Requirements for Water Ecological Monitoring-Freshwater phytoplankton "《HJ1216-2021Determination of phytoplankton in water quality0.1mLCounting box-Microscope Counting Method "and《HJ1215-2021Water quality - Determination of phytoplankton - Filter membrane-Microscopic Counting Method "," Monitoring and Analysis Methods for Water and Wastewater "(Fourth Edition), and《GB17378-2007The algae analysis requirements corresponding to the Marine Monitoring Standards.

2. Fully automatic scanning microscopy system

Ø Rack: OlympusBX43Microscope;4、10、40Double flat field achromatic objective lens20Double flat field apochromatic objective lens

Ø High precision electronic controlXYZAutomatic scanning platform: implementationX/Y/ZMicro scale axial motion and automatic control

Ø stepper motorXYPlatform: One loading4Piece, minimum step size≤ 0.1umBi directional repetitive positioning accuracy≤±1umMaximum speed:20mm/s

Ø According to the adjusted density of phytoplankton in the sample, imaging can be performed using methods such as whole sheet scanning, grid scanning, and random field scanning

Ø electricZaxisClosed loop resolution0.156umRepetitive positioning accuracy:≤±0.4um

Ø High sensitivity global shutter camera, multi depth continuous automatic scanning focus, adjustable shooting layer spacing, image resolution<0.20um/pixel

3. AlgaeAI 700 Fast counting plankton based on super deep learningAIsystem

Quick numbersAlgaeAI 700planktonAIThe automatic classification and counting system, developed by a team of senior experts, is based on in-depth research on the characteristics of plankton and combined with machine learning theory to innovatively establish a highly robust artificial intelligence analysis system. It achieves automatic classification and counting, size measurement, and biomass determination of algae and plankton.

Ø Automatically recognizable3～1000μmAlgae of various phyla, including Chlorophyta, Cyanobacteria, Diatomia, Cryptophyta, Cyanobacteria, Chlorella, Chrysophyta, Naked Algae, etc145Common algae and algae density detection range9.2×10² -10¹¹ cells/L

Ø Single field automatic recognition and analysis time≤0.6In seconds, accurate algae identification, classification and counting can be achieved, and dominant algae sorting and biomass calculation can be synchronously completed.

Ø The identification rate of dominant species in the local classification recognition library≥95%The repeatability error of automatic analysis≤5%

Ø One click operation, full process dynamic visualization: main window images line up and move rapidly, algae cells are instantly recognized, and names are labeled in situ; Real time fluctuation updates of detection data (category, name, quantity, percentage, algal density, etc.); The green scrollbar displays the progress of image collection detection. Transparent operation throughout the process, facilitating quality monitoring. Mouse interaction allows for the addition, deletion, and modification of recognized species information, and real-time updates of sample analysis results.

Ø Sort the statistical data by dominant species, displaying the category, Chinese name, Latin name, number, proportion, and density of phytoplankton. Calculate the average single-cell length, width, height, diameter, area, and volume of each species, and automatically calculate biomass and total biomassShannonIndex, species evenness index, biodiversity index, abundance, dominance.

Ø Electronic Records, Data Traceability, and Reporting: Automatically save data and generate statistical reports with just one click. The completed analysis results can be saved, and the algae names can be marked in situ on the collected images. At any time, the statistical accuracy of each image can be reviewed again by opening the file.

Ø High robustness: It has strong anti-interference ability. For field of view images containing a large amount of impurities, even if algae cells are in the impurities, this system can accurately identify them based on the inference ability of *.

Ø overlap/Adhesive algaeSeparation and identification: for algae cells that overlap highly together,AlgaeAI 700Based on intelligent adhesion separation technology, it is possible to accurately capture individual algal cells from a pile of adhered cells.

Ø incomplete/Local algaeIntelligent recognition: for incomplete algal cells at the edge of the field of view,AlgaeAI 700Based on intelligent morphological reasoning technology, it is possible to accurately identify which algae it is based on local information, thereby achieving leak free detection.

Ø Fuzzy cellCalculation recognition: for algae cells that appear light and unclear in the field of view due to insufficient depth of focus,AlgaeAI 700Based on fuzzy inference technology, it is possible to accurately analyze and identify which algal cell it is.

Ø Zooplankton analysis module: relying on *AIImage recognition technology, constructing high-precision neural network mathematical models, accurately identifying water bodies65Automatically measure indicators such as length and width of planktonic animals, calculate density and biomass, issue detection reports, and achieve paperless data recording for each species. In addition, the system provides an information database that includes text, hand drawn images, and microscopic photographs. It is equipped with classification information and keyword search functions, which can display zooplankton with both text and images.

4. Classic Image Segmentation Counting Module

Ø Dynamic automatic counting: Seven segmentation algorithms used for pre checking multi view counting and adjusting phytoplankton density to10^7-10⁸a/rise

Ø Spherical like colony cell automatic counting: automatic recognition and counting of daughter cells in the colony, especially suitable for counting and analysis of Microcystis aeruginosa

Ø Estimation of filamentous cells: used to estimate the number of daughter cells in a single filamentous or linear body

5. Qualitative analysis and intelligent identification module for planktonic organisms

Ø Plankton Expert Database: Composed of exquisite color micrographs, hand drawn images, bilingual displays in Chinese and Latin, it forms freshwater and marine plankton pools that can be accessed by“Door, order, genus, species”Four level expansion search. among whichalgae15A door1700Individual genus;zooplankton26Major categories2000Individual belonging. Freshwater algae covering the plain lake area in eastern China, the Yunnan-Guizhou Plateau lake area, Northeast lake area, Qinghai Tibet Plateau lake area, Mongolian Xinjiang Plateau lake area and seven major water systems, as well as marine algae around the East China Sea, Yellow Sea, Bohai Sea and South China Sea

Ø Typical combination association "morphological retrieval: using graphic language, combination association, and combining the structural characteristics of cells or populations to achieve accurate and fast morphological retrieval. Equipped with features such as multiple selection, freshwater and ocean storage, and easy browsing, beginners can quickly master them.

Ø Multidimensional progressive similarity algae search and identification: an automatic and intelligent algae cell graphic recognition tool,3-5Instant implementation: detecting unknown algal cell contours, extracting feature information, matching big data, accurately identifying possible algae with similar morphology, and synchronously displaying the most similar common algae through the "priority selection" option.

Ø Identification of easily confused algae: Designed for inexperienced experimenters, multiple algae that are easily confused due to their similar morphology are screened and quickly compared on the same interface. Through typical feature puzzles and summary text, the distinguishing points are quickly grasped.

6. Configuration List

Ø Quick numbersAlgaeAI700Intelligent analysis systems for algae and planktonic animals1set

Ø Fully automatic digital microscopy imaging scanning system1Set:Rack OlympusBX43Microscope,4XThe10XThe40XFlat field achromatic objective lens20XFlat field apochromatic objective lens10Adjustable diopter eyepiece, trinocular observation tube5Hole position objective lens converter,4High precision electronic control for sheet fluxXYZAutomatic scanning platform and control box, high-sensitivity global scanning camera

Ø Data analysis workstation1Stage: The12Intelligent Intel Corei9-12900 16Nuclear,32G DDR4Memory,4GIndependent graphics card,512GSolid state drives,4THard drive,27Inch display,Windows 10Professional version operating system

7. service

Ø New machine on-site installation, debugging, training, and two-year warranty service provided

Ø Build a free local database algorithm for users once

Ø Long term provision of remote assistance and guidance servicesAssist in identifying complex samples