cars 1.0.0rc1__cp313-cp313-musllinux_1_2_i686.whl

cars/pipelines/parameters/sensor_inputs.py

@@ -0,0 +1,842 @@
+#!/usr/bin/env python
+# coding: utf8
+#
+# Copyright (c) 2020 Centre National d'Etudes Spatiales (CNES).
+#
+# This file is part of CARS
+# (see https://github.com/CNES/cars).
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# pylint: disable=C0302
+"""
+CARS containing inputs checking for sensor input data
+Used for full_res and low_res pipelines
+"""
+
+import logging
+import math
+import os
+
+import numpy as np
+import rasterio as rio
+from json_checker import Checker, Or
+
+# CARS imports
+from cars.core import inputs, projection
+from cars.core.geometry.abstract_geometry import AbstractGeometry
+from cars.core.utils import make_relative_path_absolute
+from cars.orchestrator.cluster.log_wrapper import cars_profile
+from cars.pipelines.parameters import sensor_inputs_constants as sens_cst
+from cars.pipelines.parameters.sensor_loaders.sensor_loader import SensorLoader
+
+CARS_GEOID_PATH = "geoid/egm96.grd"  # Path in cars package (pkg)
+
+
+def sensors_check_inputs(conf, config_dir=None):  # noqa: C901
+    """
+    Check the inputs given
+
+    :param conf: configuration of inputs
+    :type conf: dict
+    :param config_dir: path to dir containing json
+    :type config_dir: str
+    """
+
+    overloaded_conf = conf.copy()
+
+    overloaded_conf[sens_cst.ROI] = conf.get(sens_cst.ROI, None)
+
+    overloaded_conf[sens_cst.PAIRING] = conf.get(sens_cst.PAIRING, None)
+
+    overloaded_conf[sens_cst.INITIAL_ELEVATION] = get_initial_elevation(
+        conf.get(sens_cst.INITIAL_ELEVATION, None)
+    )
+
+    overloaded_conf[sens_cst.LOADERS] = check_loaders(
+        conf.get(sens_cst.LOADERS, {})
+    )
+
+    classif_loader = overloaded_conf[sens_cst.LOADERS][
+        sens_cst.INPUT_CLASSIFICATION
+    ]
+
+    overloaded_conf[sens_cst.FILLING] = check_filling(
+        conf.get(sens_cst.FILLING, {}), classif_loader
+    )
+
+    # Validate inputs
+    inputs_schema = {
+        sens_cst.SENSORS: dict,
+        sens_cst.PAIRING: Or([[str]], None),
+        sens_cst.INITIAL_ELEVATION: Or(str, dict, None),
+        sens_cst.ROI: Or(str, dict, None),
+        sens_cst.LOADERS: dict,
+        sens_cst.FILLING: dict,
+    }
+
+    checker_inputs = Checker(inputs_schema)
+    checker_inputs.validate(overloaded_conf)
+
+    check_sensors(conf, overloaded_conf, config_dir)
+
+    # Check srtm dir
+    check_srtm(overloaded_conf[sens_cst.INITIAL_ELEVATION][sens_cst.DEM_PATH])
+
+    return overloaded_conf
+
+
+def check_sensors(conf, overloaded_conf, config_dir=None):  # noqa: C901
+    """
+    Check sensors
+    """
+    # Validate each sensor image
+    sensor_schema = {
+        sens_cst.INPUT_IMG: Or(str, dict),
+        sens_cst.INPUT_GEO_MODEL: Or(str, dict),
+        sens_cst.INPUT_MSK: Or(str, None),
+        sens_cst.INPUT_CLASSIFICATION: Or(str, dict, None),
+    }
+
+    checker_sensor = Checker(sensor_schema)
+
+    for sensor_image_key in conf[sens_cst.SENSORS]:
+        # Case where the sensor is defined as a string refering to the input
+        # image instead of a dict
+        if isinstance(conf[sens_cst.SENSORS][sensor_image_key], str):
+            # initialize sensor dictionary
+            overloaded_conf[sens_cst.SENSORS][sensor_image_key] = {
+                sens_cst.INPUT_IMG: conf[sens_cst.SENSORS][sensor_image_key]
+            }
+
+        # Overload parameters
+        image = overloaded_conf[sens_cst.SENSORS][sensor_image_key].get(
+            sens_cst.INPUT_IMG, None
+        )
+        loader_name = (
+            overloaded_conf[sens_cst.LOADERS][sens_cst.INPUT_IMG]
+            + "_"
+            + sens_cst.INPUT_IMG
+        )
+        image_loader = SensorLoader(loader_name, image, config_dir)
+        image_as_pivot_format = (
+            image_loader.get_pivot_format()  # pylint: disable=E1101
+        )
+        overloaded_conf[sens_cst.SENSORS][sensor_image_key][
+            sens_cst.INPUT_IMG
+        ] = image_as_pivot_format
+        image_path = image_as_pivot_format["bands"]["b0"]["path"]
+
+        geomodel = overloaded_conf[sens_cst.SENSORS][sensor_image_key].get(
+            "geomodel",
+            image_path,
+        )
+        overloaded_conf[sens_cst.SENSORS][sensor_image_key][
+            "geomodel"
+        ] = geomodel
+
+        mask = overloaded_conf[sens_cst.SENSORS][sensor_image_key].get(
+            sens_cst.INPUT_MSK, None
+        )
+        overloaded_conf[sens_cst.SENSORS][sensor_image_key][
+            sens_cst.INPUT_MSK
+        ] = mask
+
+        classif = overloaded_conf[sens_cst.SENSORS][sensor_image_key].get(
+            sens_cst.INPUT_CLASSIFICATION, None
+        )
+        if classif is not None:
+            loader_name = (
+                overloaded_conf[sens_cst.LOADERS][sens_cst.INPUT_CLASSIFICATION]
+                + "_"
+                + sens_cst.INPUT_CLASSIFICATION
+            )
+            classif_loader = SensorLoader(loader_name, classif, config_dir)
+            classif_as_pivot_format = (
+                classif_loader.get_pivot_format()  # pylint: disable=E1101
+            )
+            overloaded_conf[sens_cst.SENSORS][sensor_image_key][
+                sens_cst.INPUT_CLASSIFICATION
+            ] = classif_as_pivot_format
+        else:
+            overloaded_conf[sens_cst.SENSORS][sensor_image_key][
+                sens_cst.INPUT_CLASSIFICATION
+            ] = None
+
+        # Validate
+        checker_sensor.validate(
+            overloaded_conf[sens_cst.SENSORS][sensor_image_key]
+        )
+
+    # Image are now in pivot format
+    overloaded_conf[sens_cst.LOADERS][sens_cst.INPUT_IMG] = "pivot"
+    overloaded_conf[sens_cst.LOADERS][sens_cst.INPUT_CLASSIFICATION] = "pivot"
+
+    # Modify to absolute path
+    if config_dir is not None:
+        modify_to_absolute_path(config_dir, overloaded_conf)
+
+    # Check image, msk and color size compatibility
+    for sensor_image_key in overloaded_conf[sens_cst.SENSORS]:
+        sensor_image = overloaded_conf[sens_cst.SENSORS][sensor_image_key]
+        check_input_size(
+            sensor_image[sens_cst.INPUT_IMG],
+            sensor_image[sens_cst.INPUT_MSK],
+            sensor_image[sens_cst.INPUT_CLASSIFICATION],
+        )
+        # check band nbits of msk
+        check_nbits(
+            sensor_image[sens_cst.INPUT_MSK],
+        )
+
+    # Validate pairs
+    # If there is two inputs with no associated pairing, consider that the first
+    # image is left and the second image is right
+    if (
+        overloaded_conf[sens_cst.PAIRING] is None
+        and len(overloaded_conf[sens_cst.SENSORS]) == 2
+    ):
+        sensor_keys = list(overloaded_conf[sens_cst.SENSORS].keys())
+        overloaded_conf[sens_cst.PAIRING] = [[sensor_keys[0], sensor_keys[1]]]
+        logging.info(
+            (
+                "Pairing is not defined, '{}' will be used as left sensor and "
+                + "'{}' will be used as right sensor"
+            ).format(sensor_keys[0], sensor_keys[1])
+        )
+
+    if overloaded_conf[sens_cst.PAIRING] is None:
+        raise RuntimeError(
+            "Pairing is not defined and cannot be determined "
+            + "because there are more than two inputs products"
+        )
+
+    for key1, key2 in overloaded_conf[sens_cst.PAIRING]:
+        if key1 not in overloaded_conf[sens_cst.SENSORS]:
+            logging.error("{} not in sensors images".format(key1))
+            raise RuntimeError("{} not in sensors images".format(key1))
+        if key2 not in overloaded_conf["sensors"]:
+            logging.error("{} not in sensors images".format(key2))
+            raise RuntimeError("{} not in sensors images".format(key2))
+
+    # Modify to absolute path
+    if config_dir is not None:
+        modify_to_absolute_path(config_dir, overloaded_conf)
+    else:
+        logging.debug(
+            "path of config file was not given,"
+            "relative path are not transformed to absolute paths"
+        )
+
+    # Check consistency of pairs images
+    for key1, key2 in overloaded_conf[sens_cst.PAIRING]:
+        compare_image_type(
+            overloaded_conf[sens_cst.SENSORS], sens_cst.INPUT_IMG, key1, key2
+        )
+        compare_classification_values(
+            overloaded_conf[sens_cst.SENSORS],
+            sens_cst.INPUT_CLASSIFICATION,
+            key1,
+            key2,
+            overloaded_conf[sens_cst.FILLING],
+        )
+
+    return overloaded_conf
+
+
+def check_loaders(conf):
+    """
+    Check loaders section
+    :param conf: loaders section of input conf
+    :type conf: dict
+    """
+    overloaded_conf = conf.copy()
+    overloaded_conf[sens_cst.INPUT_IMG] = conf.get(sens_cst.INPUT_IMG, "basic")
+    overloaded_conf[sens_cst.INPUT_CLASSIFICATION] = conf.get(
+        sens_cst.INPUT_CLASSIFICATION, "basic"
+    )
+
+    # Validate loaders
+    loaders_schema = {
+        sens_cst.INPUT_IMG: str,
+        sens_cst.INPUT_CLASSIFICATION: str,
+    }
+
+    checker_loaders = Checker(loaders_schema)
+    checker_loaders.validate(overloaded_conf)
+
+    return overloaded_conf
+
+
+def check_filling(conf, classif_loader):
+    """
+    Check filling section
+    :param conf: filling section of input conf
+    :type conf: dict
+    """
+    basic_filling = {
+        "fill_with_geoid": None,
+        "interpolate_from_borders": None,
+        "fill_with_endogenous_dem": None,
+        "fill_with_exogenous_dem": None,
+    }
+    slurp_filling = {
+        "fill_with_geoid": [8],
+        "interpolate_from_borders": [9],
+        "fill_with_endogenous_dem": [10],
+        "fill_with_exogenous_dem": [6],
+    }
+    filling_from_loader = {}
+    filling_from_loader["basic"] = basic_filling
+    filling_from_loader["slurp"] = slurp_filling
+    filling_from_loader["pivot"] = basic_filling
+    default_filling = filling_from_loader[classif_loader]
+    overloaded_conf = conf.copy()
+    for filling_method in basic_filling:
+        overloaded_conf[filling_method] = conf.get(
+            filling_method, default_filling[filling_method]
+        )
+        if isinstance(overloaded_conf[filling_method], int):
+            overloaded_conf[filling_method] = [overloaded_conf[filling_method]]
+
+    # Validate loaders
+    loaders_schema = {
+        "fill_with_geoid": Or(None, [int]),
+        "interpolate_from_borders": Or(None, [int]),
+        "fill_with_endogenous_dem": Or(None, [int]),
+        "fill_with_exogenous_dem": Or(None, [int]),
+    }
+
+    checker_loaders = Checker(loaders_schema)
+    checker_loaders.validate(overloaded_conf)
+
+    return overloaded_conf
+
+
+def get_sensor_resolution(
+    geom_plugin, sensor_path, geomodel, target_epsg=32631
+):
+    """
+    Estimate the sensor image resolution in meters per pixel
+    using geolocation of 3 corners of the image.
+
+    :param geom_plugin: geometry plugin instance
+    :param sensor_path: path to the sensor image
+    :type sensor_path: dict
+    :param geomodel: geometric model for the sensor image
+    :param target_epsg: target EPSG code for projection
+    :type target_epsg: int
+    :return: average resolution in meters/pixel along x and y
+    :rtype: float
+    """
+    width, height = inputs.rasterio_get_size(sensor_path["bands"]["b0"]["path"])
+
+    upper_left = (0.5, 0.5)
+    upper_right = (width - 0.5, 0.5)
+    bottom_left = (0.5, height - 0.5)
+
+    # get geodetic coordinates
+    lat_ul, lon_ul, _ = geom_plugin.direct_loc(
+        sensor_path["bands"]["b0"]["path"],
+        geomodel,
+        np.array([upper_left[0]]),
+        np.array([upper_left[1]]),
+    )
+    lat_ur, lon_ur, _ = geom_plugin.direct_loc(
+        sensor_path["bands"]["b0"]["path"],
+        geomodel,
+        np.array([upper_right[0]]),
+        np.array([upper_right[1]]),
+    )
+    lat_bl, lon_bl, _ = geom_plugin.direct_loc(
+        sensor_path["bands"]["b0"]["path"],
+        geomodel,
+        np.array([bottom_left[0]]),
+        np.array([bottom_left[1]]),
+    )
+
+    coords_ll = np.array(
+        [[lon_ul, lat_ul, 0], [lon_ur, lat_ur, 0], [lon_bl, lat_bl, 0]]
+    )
+
+    # Convert to target CRS
+    coords_xy = projection.point_cloud_conversion(coords_ll, 4326, target_epsg)
+
+    diff_x = np.linalg.norm(coords_xy[1] - coords_xy[0])  # UL to UR (width)
+    diff_y = np.linalg.norm(coords_xy[2] - coords_xy[0])  # UL to BL (height)
+
+    # resolution in meters per pixel
+    res_x = diff_x / (width - 1)
+    res_y = diff_y / (height - 1)
+
+    return (res_x + res_y) / 2
+
+
+def check_geometry_plugin(
+    conf_inputs, conf_geom_plugin, epipolar_resolution, output_dem_dir
+):
+    """
+    Check the geometry plugin with inputs
+
+    :param conf_inputs: checked configuration of inputs
+    :type conf_inputs: type
+    :param conf_advanced: checked configuration of advanced
+    :type conf_advanced: type
+    :param conf_geom_plugin: name of geometry plugin
+    :type conf_geom_plugin: str
+    :param epipolar_resolution: epipolar resolution
+    :type epipolar_resolution: int
+    :return: overload inputs conf
+             overloaded geometry plugin conf
+             geometry plugin without dem
+             geometry plugin with dem
+    """
+    if conf_geom_plugin is None:
+        conf_geom_plugin = "SharelocGeometry"
+
+    # Initialize a temporary plugin, to get the product's resolution
+    temp_geom_plugin = (
+        AbstractGeometry(  # pylint: disable=abstract-class-instantiated
+            conf_geom_plugin,
+            default_alt=sens_cst.CARS_DEFAULT_ALT,
+        )
+    )
+    average_sensor_resolution = 0
+    for _, sensor_image in conf_inputs[sens_cst.SENSORS].items():
+        sensor = sensor_image[sens_cst.INPUT_IMG]
+        geomodel = sensor_image[sens_cst.INPUT_GEO_MODEL]
+        (
+            sensor,
+            geomodel,
+        ) = temp_geom_plugin.check_product_consistency(sensor, geomodel)
+        average_sensor_resolution += (
+            get_sensor_resolution(temp_geom_plugin, sensor, geomodel)
+            * epipolar_resolution
+        )
+    average_sensor_resolution /= len(conf_inputs[sens_cst.SENSORS])
+    # approximate resolution to the highest digit:
+    #  0.47 -> 0.5
+    #  7.52 -> 8
+    #  12.9 -> 10
+    nb_digits = int(math.floor(math.log10(abs(average_sensor_resolution))))
+    scaling_coeff = round(average_sensor_resolution, -nb_digits)
+    # make it so 0.5 (CO3D) is the baseline for parameters
+    scaling_coeff *= 2
+
+    # Initialize the desired geometry plugin without elevation information
+    geom_plugin_without_dem_and_geoid = (
+        AbstractGeometry(  # pylint: disable=abstract-class-instantiated
+            conf_geom_plugin,
+            default_alt=sens_cst.CARS_DEFAULT_ALT,
+            scaling_coeff=scaling_coeff,
+        )
+    )
+
+    # Check products consistency with this plugin
+    overloaded_conf_inputs = conf_inputs.copy()
+    for sensor_key, sensor_image in conf_inputs[sens_cst.SENSORS].items():
+        sensor = sensor_image[sens_cst.INPUT_IMG]
+        geomodel = sensor_image[sens_cst.INPUT_GEO_MODEL]
+        (
+            sensor,
+            geomodel,
+        ) = geom_plugin_without_dem_and_geoid.check_product_consistency(
+            sensor, geomodel
+        )
+        overloaded_conf_inputs[sens_cst.SENSORS][sensor_key][
+            sens_cst.INPUT_IMG
+        ] = sensor
+        overloaded_conf_inputs[sens_cst.SENSORS][sensor_key][
+            sens_cst.INPUT_GEO_MODEL
+        ] = geomodel
+
+    geom_plugin_with_dem_and_geoid = generate_geometry_plugin_with_dem(
+        conf_geom_plugin,
+        conf_inputs,
+        scaling_coeff=scaling_coeff,
+        output_dem_dir=output_dem_dir,
+    )
+
+    return (
+        overloaded_conf_inputs,
+        conf_geom_plugin,
+        geom_plugin_without_dem_and_geoid,
+        geom_plugin_with_dem_and_geoid,
+        scaling_coeff,
+    )
+
+
+# pylint: disable=too-many-positional-arguments
+def generate_geometry_plugin_with_dem(
+    conf_geom_plugin,
+    conf_inputs,
+    dem=None,
+    crop_dem=True,
+    output_dem_dir=None,
+    scaling_coeff=1,
+):
+    """
+    Generate geometry plugin with dem and geoid
+
+    :param conf_geom_plugin: plugin configuration
+    :param conf_inputs: inputs configuration
+    :param dem: dem to overide the one in inputs
+    :param scaling_coeff: scaling factor for resolution
+    :type scaling_coeff: float
+
+    :return: geometry plugin object, with a dem
+    """
+
+    dem_path = (
+        dem
+        if dem is not None
+        else conf_inputs[sens_cst.INITIAL_ELEVATION][sens_cst.DEM_PATH]
+    )
+
+    if crop_dem:
+        # Get image pairs for DEM intersection with ROI
+        pairs_for_roi = []
+        for key1, key2 in conf_inputs[sens_cst.PAIRING]:
+            sensor1 = conf_inputs[sens_cst.SENSORS][key1]
+            sensor2 = conf_inputs[sens_cst.SENSORS][key2]
+            image1 = sensor1[sens_cst.INPUT_IMG]
+            image2 = sensor2[sens_cst.INPUT_IMG]
+            geomodel1 = sensor1[sens_cst.INPUT_GEO_MODEL]
+            geomodel2 = sensor2[sens_cst.INPUT_GEO_MODEL]
+            pairs_for_roi.append((image1, geomodel1, image2, geomodel2))
+    else:
+        pairs_for_roi = None
+
+    # Initialize a second geometry plugin with elevation information
+
+    geom_plugin_with_dem_and_geoid = (
+        AbstractGeometry(  # pylint: disable=abstract-class-instantiated
+            conf_geom_plugin,
+            dem=dem_path,
+            geoid=conf_inputs[sens_cst.INITIAL_ELEVATION][sens_cst.GEOID],
+            default_alt=sens_cst.CARS_DEFAULT_ALT,
+            pairs_for_roi=pairs_for_roi,
+            scaling_coeff=scaling_coeff,
+            output_dem_dir=output_dem_dir,
+        )
+    )
+
+    return geom_plugin_with_dem_and_geoid
+
+
+def modify_to_absolute_path(config_dir, overloaded_conf):
+    """
+    Modify input file path to absolute path
+
+    :param config_dir: directory of the json configuration
+    :type config_dir: str
+    :param overloaded_conf: overloaded configuration json
+    :dict overloaded_conf: dict
+    """
+
+    for sensor_image_key in overloaded_conf[sens_cst.SENSORS]:
+        sensor_image = overloaded_conf[sens_cst.SENSORS][sensor_image_key]
+        for tag in [
+            sens_cst.INPUT_MSK,
+            sens_cst.INPUT_GEO_MODEL,
+        ]:
+            if isinstance(sensor_image[tag], dict):
+                sensor_image[tag][sens_cst.INPUT_PATH] = (
+                    make_relative_path_absolute(
+                        sensor_image[tag][sens_cst.INPUT_PATH], config_dir
+                    )
+                )
+            elif sensor_image[tag] is not None:
+                sensor_image[tag] = make_relative_path_absolute(
+                    sensor_image[tag], config_dir
+                )
+
+    if overloaded_conf[sens_cst.ROI] is not None:
+        if isinstance(overloaded_conf[sens_cst.ROI], str):
+            overloaded_conf[sens_cst.ROI] = make_relative_path_absolute(
+                overloaded_conf[sens_cst.ROI], config_dir
+            )
+
+    for tag in [sens_cst.DEM_PATH, sens_cst.GEOID]:
+        if overloaded_conf[sens_cst.INITIAL_ELEVATION][tag] is not None:
+            if isinstance(
+                overloaded_conf[sens_cst.INITIAL_ELEVATION][tag], str
+            ):
+                overloaded_conf[sens_cst.INITIAL_ELEVATION][tag] = (
+                    make_relative_path_absolute(
+                        overloaded_conf[sens_cst.INITIAL_ELEVATION][tag],
+                        config_dir,
+                    )
+                )
+
+
+def check_srtm(srtm_dir):
+    """
+    Check srtm data
+
+    :param srtm_dir: directory of srtm
+    :type srtm_dir: str
+    """
+
+    if srtm_dir is not None:
+        if os.path.isdir(srtm_dir):
+            srtm_tiles = os.listdir(srtm_dir)
+            if len(srtm_tiles) == 0:
+                logging.warning(
+                    "SRTM directory is empty, "
+                    "the default altitude will be used as reference altitude."
+                )
+            else:
+                logging.info(
+                    "Indicated SRTM tiles valid regions "
+                    "will be used as reference altitudes "
+                    "(the default altitude is used "
+                    "for undefined regions of the SRTM)"
+                )
+        else:
+            # TODO add check for single file
+            pass
+    else:
+        logging.info("The default altitude will be used as reference altitude.")
+
+
+def check_input_data(image, color):
+    """
+    Check data of the image and color
+
+    :param image: image path
+    :type image: str
+    :param color: color path
+    :type color: str
+    """
+
+    with rio.open(image) as img_reader:
+        trans = img_reader.transform
+        if trans.e < 0:
+            logging.warning(
+                "{} seems to have an incoherent pixel size. "
+                "Input images has to be in sensor geometry.".format(image)
+            )
+
+    with rio.open(color) as img_reader:
+        trans = img_reader.transform
+        if trans.e < 0:
+            logging.warning(
+                "{} seems to have an incoherent pixel size. "
+                "Input images has to be in sensor geometry.".format(image)
+            )
+
+
+def get_initial_elevation(config):
+    """
+    Return initial elevation parameters (dem and geoid paths)
+    from input configuration.
+
+    :param config: input initial elevation
+    :type config: str, dict or None
+    """
+
+    # Case 1 config is already a dict
+    if isinstance(config, dict):
+        updated_config = config
+    else:
+        updated_config = {}
+        updated_config[sens_cst.DEM_PATH] = (
+            config if isinstance(config, str) else None
+        )
+
+    # Add geoid path to the initial_elevation dict
+    if sens_cst.GEOID not in updated_config:
+        # use cars geoid
+        logging.info("CARS will use its own internal file as geoid reference")
+        # Get root package directory
+        package_path = os.path.dirname(__file__)
+        geoid_path = os.path.join(
+            package_path, "..", "..", "conf", CARS_GEOID_PATH
+        )
+        updated_config[sens_cst.GEOID] = geoid_path
+
+    return updated_config
+
+
+def check_input_size(image, mask, classif):
+    """
+    Check image, mask, classif and color given
+
+    Images must have same size
+
+    :param image: image path
+    :type image: str
+    :param mask: mask path
+    :type mask: str
+    :param color: color path
+    :type color: str
+    :param classif: classif path
+    :type classif: str
+    """
+    image = image["bands"]["b0"]["path"]
+    if classif is not None:
+        classif = classif[sens_cst.INPUT_PATH]
+
+    if mask is not None:
+        if inputs.rasterio_get_size(image) != inputs.rasterio_get_size(mask):
+            raise RuntimeError(
+                "The image {} and the mask {} "
+                "do not have the same size".format(image, mask)
+            )
+
+    if classif is not None:
+        if inputs.rasterio_get_size(image) != inputs.rasterio_get_size(classif):
+            raise RuntimeError(
+                "The classification bands {} and {} "
+                "do not have the same size".format(image, classif)
+            )
+
+
+def check_nbits(mask):
+    """
+    Check the bits number of the mask, classif
+    mask and classification are limited to 1 bits per band
+
+    :param mask: mask path
+    :type mask: str
+    :param classif: classif path
+    :type classif: str
+    """
+    if mask is not None:
+        nbits = inputs.rasterio_get_nbits(mask)
+        if not check_all_nbits_equal_one(nbits):
+            raise RuntimeError(
+                "The mask {} have {} nbits per band. ".format(mask, nbits)
+                + "Only the mask with nbits=1 is supported! "
+            )
+
+
+def compare_image_type(sensors, sensor_type, key1, key2):
+    """
+    Compare the data type between a pair of images
+
+    :param sensors: list of  sensor paths
+    :type sensors: str
+    :param sensor_type: type of cardataset image (IMG, MASK, CLASSIF...)
+    :type sensor_type: int
+    :param key1: key of the images pair
+    :type key1: str
+    :param key2: other key of the images pair
+    :type key2: str
+    """
+    dtype1 = inputs.rasterio_get_image_type(
+        sensors[key1][sensor_type]["bands"]["b0"]["path"]
+    )
+    dtype2 = inputs.rasterio_get_image_type(
+        sensors[key2][sensor_type]["bands"]["b0"]["path"]
+    )
+
+    if dtype1 != dtype2:
+        raise RuntimeError(
+            "The pair images haven't the same data type."
+            + "\nSensor[{}]: {}".format(key1, dtype1)
+            + "; Sensor[{}]: {}".format(key2, dtype2)
+        )
+
+
+def compare_classification_values(sensors, sensor_type, key1, key2, filling):
+    """
+    Compare the classification values between a pair of images
+
+    :param imgs: list of image paths
+    :type imgs: str
+    :param classif_type: type of cardataset image (IMG, MASK, CLASSIF...)
+    :type classif_type: int
+    :param key1: key of the images pair
+    :type key1: str
+    :param key2: other key of the images pair
+    :type key2: str
+    """
+    classif1 = sensors[key1][sensor_type]
+    classif2 = sensors[key2][sensor_type]
+    if classif1 is not None and classif2 is not None:
+        values1 = classif1[sens_cst.INPUT_VALUES]
+        values2 = classif2[sens_cst.INPUT_VALUES]
+        all_values = list(set(values1) | set(values2))
+        classif1[sens_cst.INPUT_VALUES] = all_values
+        classif2[sens_cst.INPUT_VALUES] = all_values
+        for filling_method in filling:
+            filling_values = filling[filling_method]
+            if filling_values is not None and not all(
+                isinstance(val, int) for val in filling_values
+            ):
+                raise TypeError(
+                    "Not all values defined for "
+                    "filling {} are int : {}".format(
+                        filling_method,
+                        filling_values,
+                    )
+                )
+            if filling_values is not None and not set(filling_values) <= set(
+                all_values
+            ):
+                logging.warning(
+                    "One of the values {} on which filling {} must be applied "
+                    "does not exist on classifications {} and {}".format(
+                        filling_values,
+                        filling_method,
+                        classif1[sens_cst.INPUT_PATH],
+                        classif2[sens_cst.INPUT_PATH],
+                    )
+                )
+
+
+def check_all_nbits_equal_one(nbits):
+    """
+    Check if all the nbits = 1
+    :param nbits: list of the nbits
+    :return: True if all the nbits = 1
+    """
+    if len(nbits) > 0 and nbits[0] == 1 and all(x == nbits[0] for x in nbits):
+        return True
+    return False
+
+
+@cars_profile(name="Generate inputs")
+def generate_inputs(conf, geometry_plugin):
+    """
+    Generate sensors inputs form inputs conf :
+
+    a list of (sensor_left, sensor_right)
+
+    :param conf: input conf
+    :type conf: dict
+
+    :return: list of sensors pairs
+    :rtype: list(tuple(dict, dict))
+
+    """
+    # Load geomodels directly on conf object
+    sensors = conf[sens_cst.SENSORS]
+    for key in sensors:
+        geomodel = sensors[key][sens_cst.INPUT_GEO_MODEL]
+        loaded_geomodel = geometry_plugin.load_geomodel(geomodel)
+        sensors[key][sens_cst.INPUT_GEO_MODEL] = loaded_geomodel
+
+    # Get needed pairs
+    pairs = conf[sens_cst.PAIRING]
+
+    # Generate list of pairs
+    list_sensor_pairs = []
+    for key1, key2 in pairs:
+        merged_key = key1 + "_" + key2
+        sensor1 = conf[sens_cst.SENSORS][key1]
+        sensor2 = conf[sens_cst.SENSORS][key2]
+        list_sensor_pairs.append((merged_key, sensor1, sensor2))
+
+    return list_sensor_pairs