Test data and model for the FlowCam data processing pipeline

Testing data for the processing pipeline for FlowCam data 

The data are fully processed but can be used to test each pipeline component. You can download the scripts at <github> and LabelChecker <here>

Pipeline scripts

To use the model, unzip the freshwater_phytoplankton_model.zip and place the folder in the respective model folder in the services. 

|--services
    |-- ProcessData.py
    |-- config.py
    |--  classification
        |-- ObjectClassification
            |-- models
                |-- <paste here>
            |-- ...
    |-- ...
|-- ...

Once you unzip the data.zip file, each folder corresponds to the data export of a FlowCam run. You have the TIF collage files, a CSV file with the sample name containing all the parameters measured by the FlowCam, and a LabelChecker_<sample> CSV file generated by the preprocessing.py script. 

You can run the preprocessing.py script directly on the files by including the -R (reprocess) argument. Otherwise you can do it by removing the LabelChecker CSV from the folders. The PreprocessingTrue column will remain the same.

When running the classification.py script you can get new predictions on the data. In this case, only the LabelPredicted column will be updated and the validated labels (LabelTrue column) will not be lost. 

You could also use these files to try out the train_model.ipynb, although the resulting model will not be very good with so little data. We recommend trying it with your own data. 

LabelChecker

These files can be used to test LabelChecker. You can open them one by one or all together and try all functionalities. We provide a label_file.csv but you can also make your own.