helpers

setTFConfig

This method configures the TensorFlow library settings along with configuring GPU settings.

---

printSampleCounts

This method prints the respective samples in the given dataframes.

def printSampleCounts(plotClassifications, training_dataframe, validation_dataframe, test_dataframe=None):
    if test_dataframe != None:
        print("Class | Train | Valid | Test")
    else:
        print("Class | Train | Valid")

    for Class in plotClassifications:
        className = Class["Classification"]
        trainCount = training_dataframe.loc[training_dataframe.label == className].shape[0]
        validCount = validation_dataframe.loc[validation_dataframe.label == className].shape[0]
        row=className+"  |  "+str(trainCount)+" ("+str(float(trainCount)/float(training_dataframe.shape[0]))+")  |  "+str(validCount)+" ("+str(float(validCount)/validation_dataframe.shape[0])+")"
        if test_dataframe != None:
            testCount = test_dataframe.loc[test_dataframe.label == className].shape[0]
            row+" | "+str(testCount)+" ("+str(float(testCount)/test_dataframe.shape[0])+")"
        print(row)
        # logging.info(row)
        if validCount == 0:
            print("Missing example for "+str(className)+" exiting")
            # logging.error("Missing example for "+str(className)+" exiting")

Parameters

• plotClassifications: A list representing the classifications of the plots

• training_dataframe: A Pandas DataFrame of the training data

• validation_dataframe: A Pandas DataFrame of the validation data

• test_dataframe: An optional dataframe of the test data

---

getGenerator

This method returns data generators for the validation, training, and testing datasets. The pixels are normalized and the generator is configured to respective settings. If a testing dataframe is unavailable, the validation dataframe is used to predict a possible dataframe.

# Extended code available on Github
def getGenerator(training_dataframe, validation_dataframe, test_dataframe=None, BS=32):

Parameters

• training_dataframe: A Pandas DataFrame of the training data

• validation_dataframe: A Pandas DataFrame of the validation data

• test_dataframe: An optional dataframe of the test data

• BS: The generators’ batch size. Defaults to 32.