Res50.yaml

# Deep-learning-base Ionogram Automatic Scaler Configuration File

# Data Parameters
Data:
    BasePath: ./dataset/
    TrainListFile: train.lst
    TestListFile: test.lst
    # pad input height to 512
    PadHeight: 512
    # pad input width to 512
    PadWidth: 512
    # number of input channels
    ChannelNum: 3 
    # Number of predicted classes
    ClassNum: 3
# Model Parameters
Model:
    Type: FPN
    # Backbone available params: 
    # VGG 'vgg16' 'vgg19'
    # ResNet 'resnet18' 'resnet34' 'resnet50' 'resnet101' 'resnet152'
    # SE-ResNet 'seresnet18' 'seresnet34' 'seresnet50' 'seresnet101' 'seresnet152'
    # ResNeXt 'resnext50' 'resnext101'
    # SE-ResNeXt 'seresnext50' 'seresnext101'
    # SENet154 'senet154'
    # DenseNet 'densenet121' 'densenet169' 'densenet201'
    # Inception 'inceptionv3' 'inceptionresnetv2'
    # MobileNet 'mobilenet' 'mobilenetv2'
    # EfficientNet 'efficientnetb0' 'efficientnetb1' 'efficientnetb2' 'efficientnetb3' 'efficientnetb4' 'efficientnetb5' efficientnetb6' efficientnetb7'
    Backbone: resnet50
    # BackboneWeights available params:
    # None imagenet
    BackboneWeights: None
    # Number of pyramid block filters
    PyramidBlockFilters: 256
    # optimizer
    Optimizer: Adam
    # Learning rate
    LearningRate: 0.0001
    # Loss function available params:
    # bce (Binary Cross Entropy)
    # dice
    # BinaryFocalLoss
    Loss: BinaryFocalLoss
    # Evalute metric available params:
    # IoU
    # F_score
    Metric: F_score

Train:
    # total training step
    TotalStep: 160000
    # print interval
    PrintInterval: 100
    # model save step
    SaveInterval: 2000
    # test img plot interval
    PlotInterval: 1000
    # img save dirs's
    ImgSaveDir: /home/Public/SmartRadarAssistant/DIonoAutoScaler/FPN_compare/Res50FPN/imgs/
    # model save dir
    ModelSaveDir: /home/Public/SmartRadarAssistant/DIonoAutoScaler/FPN_compare/Res50FPN/models/
    HistLogDir: /home/Public/SmartRadarAssistant/DIonoAutoScaler/FPN_compare/Res50FPN/