example_config.yaml

# ==========================
# CellViT Inference Config
# ==========================

# Model selection (REQUIRED)
model:                # REQUIRED | str: Segmentation model to use.
                      # Choices: ["SAM", "HIPT"]
# Nuclei classification taxonomy (OPTIONAL)
nuclei_taxonomy:      # OPTIONAL | str: Defines the nuclei classification taxonomy.
                      # Choices: ["binary", "pannuke", "consep", "lizard", "midog", "nucls_main", "nucls_super", "ocelot", "panoptils"]
                      # Default: "pannuke"

# ==========================
# Inference Settings (OPTIONAL)
# ==========================
inference:
  gpu:                # OPTIONAL | int: GPU ID to use for inference.
                      # Default: 0 (use first available GPU)
  enforce_amp:        # OPTIONAL | bool: Whether to use Automatic Mixed Precision (AMP) for inference.
                      # Default: false (disabled)
  batch_size:         # OPTIONAL | int: Number of images processed per batch.
                      # Default: 8

# ==========================
# Output Settings
# ==========================
output_format:
  outdir:             # REQUIRED | str: Path to the output directory where results will be stored.
  geojson:            # OPTIONAL | bool: Whether to export results in GeoJSON format (for QuPath or other tools).
                      # Default: false (disabled)
  graph:              # OPTIONAL | bool: Whether to generate a cell graph representation.
                      # Default: false (disabled)
  compression:        # OPTIONAL | bool: Whether to use Snappy compression for output files.
                      # Default: false (disabled)

# ==========================
# Processing Mode (Choose One)
# ==========================
# Either `process_wsi` (for a single image) or `process_dataset` (for multiple images) should be used.

# --- Process a Single Whole Slide Image (WSI) ---
process_wsi:
  wsi_path:           # REQUIRED | str: Path to the Whole Slide Image (WSI) file.
  wsi_mpp:            # OPTIONAL | float: Microns per pixel (spatial resolution of the slide).
                      # Default: Extracted automatically from file (if available).
  wsi_magnification:  # OPTIONAL | int: Magnification level of the slide (e.g., 20x, 40x).
                      # Default: Extracted automatically from file (if available).

# --- Process a Dataset (Multiple WSI Files) ---
process_dataset:
  wsi_folder:         # REQUIRED (if `wsi_filelist` is NOT used) | str: Path to a folder containing multiple WSI files.
                      # Either `wsi_folder` OR `wsi_filelist` must be provided (not both).
  wsi_extension:      # OPTIONAL | str: File extension of WSI files (used for detection).
                      # Default: "svs"
  wsi_filelist:       # REQUIRED (if `wsi_folder` is NOT used) | str: Path to a CSV file listing WSI files.
                      # CSV must have a 'path' column, with optional 'wsi_mpp' and 'wsi_magnification' columns.
                      # If 'wsi_mpp' and 'wsi_magnification' are provided, they override global settings.
  wsi_mpp:            # OPTIONAL | float: Microns per pixel (spatial resolution).
                      # Default: Extracted automatically from file (if available).
                      # Can be used with both `wsi_folder` and `wsi_filelist`.
  wsi_magnification:  # OPTIONAL | int: Magnification level of the slides.
                      # Default: Extracted automatically from file (if available).
                      # Can be used with both `wsi_folder` and `wsi_filelist`.

# ==========================
# System Settings (OPTIONAL)
# ==========================
system:
  cpu_count:          # OPTIONAL | int: Number of CPU cores to use for inference.
                      # Default: Uses system configuration.
  ray_worker:         # OPTIONAL | int: Number of ray worker to use for inference. Limited by cpu_count.
                      # Default: Uses system configuration.
  ray_remote_cpus:    # OPTIONAL | int: Number of cpus per ray worker.
                      # Default: Uses system configuration.
  memory:             # OPTIONAL | int: RAM in MB to use
                      # Default: Uses system configuration

# ==========================
# Debug Settings (OPTIONAL)
# ==========================
debug:                # OPTIONAL | bool: If debug should be used - this changes to logger level and requires ray[default]
                      # Default: False