parameters.yml

PIPELINES:
  __default__:
    =: pipelinex.FlexiblePipeline
    module: # None
    decorator: # None
    nodes:
      - inputs: img_source # [Tip] If `func` is omitted, `inputs` will be passed to `outputs` as is
        outputs: img_00

      - inputs: img_00
        func:
          - { =: pipelinex.CvResize, dsize: "(512, 512)" }
        outputs: img_01

      - inputs: img_01
        func:
          - =: pipelinex.SkimageSegmentationFelzenszwalb
            scale: 500
            sigma: 5
        outputs: raw_graph_seg_img

      - inputs: raw_graph_seg_img
        func:
          - { =: pipelinex.fit_to_uint8 }
        outputs: graph_seg_img

      - inputs: graph_seg_img
        decorator: { =: pipelinex.dict_io }
        func: { =: roi.seg_to_roi }
        outputs: roi

      - inputs: raw_graph_seg_img
        outputs: raw_sem_seg_img

      # - inputs: img_01
        # decorator: { =: pipelinex.dict_io }
        # func:
          # - { =: semantic_segmentation.get_semantic_segments }
        # outputs: raw_sem_seg_img

      - inputs: raw_sem_seg_img
        func:
          - { =: pipelinex.fit_to_uint8 }
        outputs: sem_seg_img

      - inputs: img_01
        func:
          - { =: pipelinex.NpZerosLike, _ }
        outputs: zero_img

      - inputs: [zero_img, raw_sem_seg_img]
        func:
          - { =: pipelinex.SkimageMarkBoundaries, mode: outer, color: 1 }
          - { =: pipelinex.NpMean, axis: 2 }
        outputs: raw_seg_edge_img

      - inputs: raw_seg_edge_img
        func:
          - { =: pipelinex.fit_to_uint8 }
        outputs: seg_edge_img

      - inputs: img_01
        func:
          #          - {=: pipelinex.CvBGR2HSV, _: }
          - { =: pipelinex.CvBGR2Gray, _ }
          - { =: pipelinex.CvEqualizeHist, _ }
          #          - {=: pipelinex.CvBlur, ksize: "(5, 5)"}
          #          - {=: pipelinex.CvMedianBlur, ksize: 5}
          #          - {=: pipelinex.CvBilateralFilter, d: 9, sigmaColor: 200, sigmaSpace: 200}
          - { =: pipelinex.CvGaussianBlur, ksize: "(5, 5)", sigmaX: 3 }
        outputs: img_02

      - inputs: img_02
        func:
          #          - {=: pipelinex.CvDiagonalEdgeFilter2d, ddepth: -1, kernel_type: 1}
          #          - {=: pipelinex.fit_to_uint8}
          - { =: pipelinex.CvCanny, threshold1: 10, threshold2: 10, apertureSize: 5, L2gradient: True }
        outputs: img_03

      - inputs: [img_03, roi, raw_seg_edge_img, img_01]
        decorator: { =: pipelinex.dict_io }
        func: { =: empty_area.detect_lines_and_estimate_empty_ratio }
        outputs: [empty_ratios_dict, img_04_raw, img_05]

      - inputs: empty_ratios_dict
        func: { =: pipelinex.NestedDictToDf, index_name: image }
        outputs: empty_ratios_df

      - inputs: img_04_raw
        decorator: { =: pipelinex.dict_io }
        func: { =: empty_area.visualize_lines_img }
        outputs: img_04

      - inputs: [img_01, graph_seg_img]
        func:
          - { =: pipelinex.SkimageMarkBoundaries, mode: outer }
          - { =: pipelinex.fit_to_uint8 }
        outputs: vis_graph_seg_edge_img

      - inputs: [img_01, sem_seg_img]
        func:
          - { =: pipelinex.SkimageMarkBoundaries, mode: outer }
          - { =: pipelinex.fit_to_uint8 }
        outputs: vis_sem_seg_edge_img

RUN_CONFIG:
  pipeline_name: __default__
  runner: SequentialRunner # Set to "ParallelRunner" to run in parallel
  only_missing: False # Set True to run only missing nodes
  tags: # None
  node_names: # None
  from_nodes: # None
  to_nodes: # None
  from_inputs: # None
  load_versions: # None

HOOKS:
  - =: pipelinex.MLflowBasicLoggerHook # Configure and log duration time for the pipeline 
    enable_mlflow: True # Enable configuring and logging to MLflow
    uri: sqlite:///mlruns/sqlite.db
    experiment_name: experiment_001
    artifact_location: ./mlruns/experiment_001
    offset_hours: 0 # Specify the offset hour (e.g. 0 for UTC/GMT +00:00) to log in MLflow

  - =: pipelinex.MLflowArtifactsLoggerHook # Log artifacts of specified file paths and dataset names
    enable_mlflow: True # Enable logging to MLflow
    filepaths_before_pipeline_run: # Optionally specify the file paths to log before pipeline is run
      - conf/base/parameters.yml
    datasets_after_node_run: # Optionally specify the dataset names to log after the node is run
      - model
    filepaths_after_pipeline_run: # None  # Optionally specify the file paths to log after pipeline is run

  - =: pipelinex.MLflowDataSetsLoggerHook # Log datasets of (list of) float, int, and str classes
    enable_mlflow: True # Enable logging to MLflow

  - =: pipelinex.MLflowTimeLoggerHook # Log duration time to run each node (task)
    enable_mlflow: True # Enable logging to MLflow

  - =: pipelinex.AddTransformersHook # Add transformers
    transformers: 
      =: pipelinex.MLflowIOTimeLoggerTransformer # Log duration time to load and save each dataset
      enable_mlflow: True

#
# Command to run mlflow server:
# $ mlflow server --host 0.0.0.0 --backend-store-uri sqlite:///mlruns/sqlite.db --default-artifact-root ./mlruns/experiment_001