def __init__(
self,
model_dir,
weights_path,
model_config=defaults.DefaultConfig(),
class_names=defaults.CLASSES,
layout_params={},
init=None
):
self.model_config = model_config
# Check that `class_names` make sense
if len(class_names) == (model_config.NUM_CLASSES - 1):
self.class_names = ["BG"] + class_names
else:
assert (
len(class_names) == model_config.NUM_CLASSES
), "Number of `class_names` must match number in `model_config`"
assert (
class_names[0] == "BG"
), "Background `BG` must be first in `class_names`"
self.class_names = class_names
# Set defaults and check validity of any new params via setter
self._layout_params = defaults.LAYOUT_PARAMS
self.layout_params = layout_params
# Build and load the saved model
print(f"Building MRCNN model from directory: {str(model_dir)}")
self.model = modellib.MaskRCNN(
mode="inference", config=self.model_config, model_dir=str(model_dir)
)
print(f"Loading model weights from file: {str(weights_path)}")
if init == "coco":
self.model.load_weights(str(weights_path), by_name=True,
exclude=["mrcnn_class_logits",
"mrcnn_bbox_fc",
"mrcnn_bbox",
"mrcnn_mask"])
elif init == "imagenet":
self.model.load_weights(self.model.get_imagenet_weights(), by_name=True)
else:
try:
self.model.load_weights(str(weights_path), by_name=True)
except ValueError as exc:
msg = (f"Error during loading pretrained weights: {exc}")
raise CorebreakoutError(msg) from None
def __init__(
self,
model_dir,
weights_path,
model_config=defaults.DefaultConfig(),
class_names=defaults.CLASSES,
layout_params={},
):
"""
Parameters
----------
model_dir : str or Path
Path to saved MRCNN model directory
weights_path : str or Path
Path to saved weights file of corresponding model
model_config : `mrcnn.Config`, optional
Instance of MRCNN configuration object, default=`defaults.DefaultConfig()`.
class_names : list(str), optional
A list of the class names for model output. Should be in same order as in
the `Dataset` object that model was trained on. Default=`defaults.CLASSES`
layout_params : dict, optional
Any layout parameters to override from default=`defaults.LAYOUT_PARAMS`.
See `docs/layout_parameters.md` for explanations and options for each parameter.
"""
self.model_config = model_config
# Check that `class_names` make sense
if len(class_names) == (model_config.NUM_CLASSES - 1):
self.class_names = ["BG"] + class_names
else:
assert (
len(class_names) == model_config.NUM_CLASSES
), "Number of `class_names` must match number in `model_config`"
assert (class_names[0] == "BG"
), "Background `BG` must be first in `class_names`"
self.class_names = class_names
# Set defaults and check validity of any new params via setter
self._layout_params = defaults.LAYOUT_PARAMS
self.layout_params = layout_params
# Build and load the saved model
print(f"Building MRCNN model from directory: {str(model_dir)}")
self.model = modellib.MaskRCNN(mode="inference",
config=self.model_config,
model_dir=str(model_dir))
print(f"Loading model weights from file: {str(weights_path)}")
self.model.load_weights(str(weights_path), by_name=True)
"""
import os
import argparse
from glob import glob
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from functools import reduce
from operator import add
from corebreakout import defaults
from corebreakout import CoreSegmenter, CoreColumn
# Change Config selection manually
model_config = defaults.DefaultConfig()
# Change class_names manually
class_names = defaults.CLASSES
# Change any non-default layout_params manually
layout_params = defaults.LAYOUT_PARAMS
parser = argparse.ArgumentParser(
description=
'Convert image directories with Mask R-CNN and save results as `CoreColumn`s.'
)
parser.add_argument(
'path',
type=str,
help="Path to directory of images (and depth information csv) to process.")