Datasets:

LvZTTTT
/

Earth-Bench-V2

import argparse

from pathlib import Path
from fastmcp import FastMCP

from tool_utils import read_image, read_image_uint8

mcp = FastMCP("perception")
parser = argparse.ArgumentParser()
parser.add_argument('--temp_dir', default='tmp/tmp', type=str)
parser.add_argument('--port', default=20003, type=int)
args, unknown = parser.parse_known_args()

TEMP_DIR = Path(args.temp_dir)
TEMP_DIR.mkdir(parents=True, exist_ok=True)


@mcp.tool(description="""
Perform threshold-based segmentation on a single-band raster image.
The function reads a raster image from the specified path, converts it to a binary mask
by applying a fixed threshold, and writes the resulting binary image to a new file.
Pixel values greater than the threshold are set to 255 (white), and values less than or
equal to the threshold are set to 0 (black).
Parameters:
    input_image_path (str): Path to the input raster image file (e.g., TIFF, PNG, JPG).
    threshold (float or int): Pixel intensity threshold used to generate the binary mask.
    output_path (str): Relative output path (under TEMP_DIR) where the result will be saved,
                        e.g., "question17/threshold_segmentation_2022-01-16.tif".
Returns:
    str: Message indicating the file path where the result is saved.
""")
def threshold_segmentation(input_image_path, threshold, output_path):
    '''
    Perform threshold-based segmentation on a single-band raster image.
    The function reads a raster image from the specified path, converts it to a binary mask
    by applying a fixed threshold, and writes the resulting binary image to a new file.
    Pixel values greater than the threshold are set to 255 (white), and values less than or
    equal to the threshold are set to 0 (black).
    Parameters:
        input_image_path (str): Path to the input raster image file (e.g., TIFF, PNG, JPG).
        threshold (float or int): Pixel intensity threshold used to generate the binary mask.
        output_path (str): Relative output path (under TEMP_DIR) where the result will be saved,
                           e.g., "question17/threshold_segmentation_2022-01-16.tif".
    Returns:
        str: Message indicating the file path where the result is saved.
    '''
    try:
        import os
        import rasterio
        import numpy as np

        with rasterio.open(input_image_path) as src:
            image = src.read(1)
            meta = src.meta.copy()

        binary_image = (image > threshold).astype(np.uint8)

        meta.update(dtype=rasterio.uint8, count=1)
        os.makedirs((TEMP_DIR / output_path).parent, exist_ok=True)
        with rasterio.open(TEMP_DIR / output_path, 'w', **meta) as dst:
            dst.write(binary_image, 1)

        return f'Result save at {TEMP_DIR / output_path}'
    except Exception as e:

        return f'Error in threshold_segmentation: {str(e)}'


@mcp.tool(description="""
Expands bounding boxes by a given radius and returns the expanded bounding boxes.
Parameters:
    bboxes (list[list[float]]): List of bounding boxes, each represented as [x1, y1, x2, y2].
    radius (float): Expansion radius in the same unit as the GSD.
    gsd (float): Ground Sampling Distance in the same unit as the radius.
Returns:
    list[list[float]]: List of expanded bounding boxes, each represented as [x1, y1, x2, y2].
""")
def bbox_expansion(bboxes: list[list[float]], radius: float, gsd: float):
    """
    Expands bounding boxes by a given radius and returns the expanded bounding boxes.
    Parameters:
        bboxes (list[list[float]]): List of bounding boxes, each represented as [x1, y1, x2, y2].
        radius (float): Expansion radius in the same unit as the GSD.
        gsd (float): Ground Sampling Distance in the same unit as the radius.
    Returns:
        list[list[float]]: List of expanded bounding boxes, each represented as [x1, y1, x2, y2].
    """
    try:
        expanded_bboxes = []
        for bbox in bboxes:
            x1, y1, x2, y2 = bbox
            x1 = x1 - radius / gsd
            y1 = y1 - radius / gsd
            x2 = x2 + radius / gsd
            y2 = y2 + radius / gsd
            expanded_bboxes.append([x1, y1, x2, y2])

        return expanded_bboxes
    except Exception as e:

        return f'Error in bbox_expansion: {str(e)}'



@mcp.tool(description="""
    Description:
        Count the number of pixels in an image whose values are greater than 
        the specified threshold.
    Parameters:
        file_path (str):
            Path to the input image (GeoTIFF or raster format).
        threshold (float):
            Threshold value for hotspot detection.
    Returns:
        count (int):
            Number of pixels with values greater than the threshold.
    Example:
        >>> count_above_threshold("sample_image.tif", 100)
        2456
    """)
def count_above_threshold(file_path: str, threshold: float):
    """
    Description:
        Count the number of pixels in an image whose values are greater than 
        the specified threshold.
    Parameters:
        file_path (str):
            Path to the input image (GeoTIFF or raster format).
        threshold (float):
            Threshold value for hotspot detection.
    Returns:
        count (int):
            Number of pixels with values greater than the threshold.
    Example:
        >>> count_above_threshold("sample_image.tif", 100)
        2456
    """
    try:
        import numpy as np
        import rasterio
        with rasterio.open(file_path) as src:
            x = src.read(1)
        x = np.asarray(x)
        # Count elements greater than threshold
        count = np.sum(x > threshold)
    
        return int(count)
    except Exception as e:

        return f'Error in count_above_threshold: {str(e)}'



@mcp.tool(description=
    """
    Description:
        Read a binary image, apply erosion and skeletonization, 
        then count the number of external contours in the skeletonized image.
    Parameters:
        image_path (str):
            Path to the input binary (black and white) image.
    Returns:
        count (int):
            Number of external contours detected after skeletonization.
    Example:
        >>> count_connected_components("binary_mask.png")
        12
    """)
def count_connected_components(image_path):
    """
    Description:
        Read a binary image, apply erosion and skeletonization, 
        then count the number of external contours in the skeletonized image.
    Parameters:
        image_path (str):
            Path to the input binary (black and white) image.
    Returns:
        count (int):
            Number of external contours detected after skeletonization.
    Example:
        >>> count_skeleton_contours("binary_mask.png")
        12
    """
    try:
        import cv2
        import numpy as np
        from skimage.morphology import skeletonize
        # Read image as grayscale
        img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)

        if img is None:
            raise FileNotFoundError(f"Failed to read image: {image_path}")

        # Binarize the image
        _, binary = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)

        # Apply erosion
        kernel = np.ones((3, 3), np.uint8)
        eroded = cv2.erode(binary, kernel, iterations=1)

        # Skeletonize
        skeleton = skeletonize(eroded > 0)  # Convert to boolean for skimage
        skeleton_uint8 = (skeleton * 255).astype(np.uint8)

        # Find contours
        contours, _ = cv2.findContours(skeleton_uint8, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

        return len(contours)
    except Exception as e:

        return f'Error in count_skeleton_contours: {str(e)}'



@mcp.tool(description=
    """
    Description:
        Convert bounding boxes from [x_min, y_min, x_max, y_max] format
        to centroid coordinates (x, y).
    Parameters:
        bboxes (list[list[float]]):
            A list of bounding boxes, each defined as [x_min, y_min, x_max, y_max].
    Returns:
        centroids (list[tuple[float, float]]):
            A list of centroid coordinates, each in (x, y) format.
    Example:
        >>> bboxes2centroids([[0, 0, 10, 20], [5, 5, 15, 15]])
        [(5.0, 10.0), (10.0, 10.0)]
    """)
def bboxes2centroids(bboxes):
    """
    Description:
        Convert bounding boxes from [x_min, y_min, x_max, y_max] format
        to centroid coordinates (x, y).
    Parameters:
        bboxes (list[list[float]]):
            A list of bounding boxes, each defined as [x_min, y_min, x_max, y_max].
    Returns:
        centroids (list[tuple[float, float]]):
            A list of centroid coordinates, each in (x, y) format.
    Example:
        >>> bboxes2centroids([[0, 0, 10, 20], [5, 5, 15, 15]])
        [(5.0, 10.0), (10.0, 10.0)]
    """
    try:
        return [((x1 + x2) / 2, (y1 + y2) / 2) for x1, y1, x2, y2 in bboxes]
    except Exception as e:

        return f'Error in bboxes2centroids: {str(e)}'



@mcp.tool(description=
    """
    Description:
        Compute pairwise distances between centroids and return both the closest 
        and farthest pairs with their indices and distances.
    Parameters:
        centroids (list[tuple[float, float]] or np.ndarray):
            A list or NumPy array of centroid coordinates in (x, y) format.
    Returns:
        result (dict):
            A dictionary containing:
              - 'min': (index1, index2, distance)
                  Indices of the closest centroid pair and their distance.
              - 'max': (index1, index2, distance)
                  Indices of the farthest centroid pair and their distance.
    Example:
        >>> centroids = [(0, 0), (3, 4), (10, 0)]
        >>> centroid_distance_extremes(centroids)
        {'min': (0, 1, 5.0), 'max': (1, 2, 7.211102550927978)}
    """)
def centroid_distance_extremes(centroids):
    """
    Description:
        Compute pairwise distances between centroids and return both the closest 
        and farthest pairs with their indices and distances.
    Parameters:
        centroids (list[tuple[float, float]] or np.ndarray):
            A list or NumPy array of centroid coordinates in (x, y) format.
    Returns:
        result (dict):
            A dictionary containing:
              - 'min': (index1, index2, distance)
                  Indices of the closest centroid pair and their distance.
              - 'max': (index1, index2, distance)
                  Indices of the farthest centroid pair and their distance.
    Example:
        >>> centroids = [(0, 0), (3, 4), (10, 0)]
        >>> centroid_distance_extremes(centroids)
        {'min': (0, 1, 5.0), 'max': (1, 2, 7.211102550927978)}
    """
    try:
        import numpy as np
        points = np.array(centroids)
        diff = points[:, None, :] - points[None, :, :]
        dist_matrix = np.sqrt(np.sum(diff ** 2, axis=-1))

        np.fill_diagonal(dist_matrix, np.inf)
        min_idx = np.unravel_index(np.argmin(dist_matrix), dist_matrix.shape)
        min_dist = dist_matrix[min_idx]

        np.fill_diagonal(dist_matrix, -np.inf)
        max_idx = np.unravel_index(np.argmax(dist_matrix), dist_matrix.shape)
        max_dist = dist_matrix[max_idx]

        return {
            "min": (int(min_idx[0]), int(min_idx[1]), float(min_dist)),
            "max": (int(max_idx[0]), int(max_idx[1]), float(max_dist))
        }
    except Exception as e:

        return f'Error in centroid_distance_extremes: {str(e)}'



@mcp.tool(description="""
    Description:
        Calculate the total area of a list of bounding boxes in [x1, y1, x2, y2] format.
    Parameters:
        bboxes (list[list[float]]):
            A list of bounding boxes, where each box is defined as [x1, y1, x2, y2].
            - x1, y1 → top-left corner coordinates
            - x2, y2 → bottom-right corner coordinates
        gsd (float, optional):
            Ground sample distance (meters per pixel).
            - If provided, the result is in square meters (m²).
            - If None, the result is in square pixels (pixel²). Default = None.
    Returns:
        total_area (float):
            The total area of all bounding boxes, in m² if gsd is provided, otherwise in pixel².
    Example:
        >>> calculate_bbox_area([[0, 0, 10, 20], [5, 5, 20, 15]])
        300.0
        >>> calculate_bbox_area([[0, 0, 10, 20]], gsd=0.5)
        50.0
    """)
def calculate_bbox_area(bboxes, gsd=None):
    """
    Description:
        Calculate the total area of a list of bounding boxes in [x1, y1, x2, y2] format.
    Parameters:
        bboxes (list[list[float]]):
            A list of bounding boxes, where each box is defined as [x1, y1, x2, y2].
            - x1, y1 → top-left corner coordinates
            - x2, y2 → bottom-right corner coordinates
        gsd (float, optional):
            Ground sample distance (meters per pixel).
            - If provided, the result is in square meters (m²).
            - If None, the result is in square pixels (pixel²). Default = None.
    Returns:
        total_area (float):
            The total area of all bounding boxes, in m² if gsd is provided, otherwise in pixel².
    Example:
        >>> calculate_bbox_area([[0, 0, 10, 20], [5, 5, 20, 15]])
        300.0
        >>> calculate_bbox_area([[0, 0, 10, 20]], gsd=0.5)
        50.0
    """
    try:
        total_area = 0.0

        for bbox in bboxes:
            if len(bbox) != 4:
                raise ValueError(f"Invalid bbox format: {bbox}. Expected [x1, y1, x2, y2].")

            x1, y1, x2, y2 = bbox

            width = max(0.0, x2 - x1)
            height = max(0.0, y2 - y1)

            area = width * height
            total_area += area

        if gsd is not None:
            total_area *= gsd * gsd

        return total_area

    except Exception as e:
        return f'Error in calculate_bbox_area: {str(e)}'

   
def bboxes_equal(bboxes1, bboxes2, diff_threshold=0.001):
    """
    Determine whether two sets of bounding boxes are equivalent by rasterizing them into masks
    and comparing pixel-wise differences.
    Args:
        bboxes1 (list of list of float): First list of bounding boxes [[x1, y1, x2, y2], ...].
        bboxes2 (list of list of float): Second list of bounding boxes.
        diff_threshold (float): Maximum allowed ratio of differing pixels.
    Returns:
        bool: True if masks are sufficiently similar, otherwise False.
    """
    import numpy as np
    if len(bboxes1) != len(bboxes2):
        return False

    all_boxes = bboxes1 + bboxes2

    max_x = max(max(b[0], b[2]) for b in all_boxes)
    max_y = max(max(b[1], b[3]) for b in all_boxes)

    W = int(np.ceil(max_x)) + 1
    H = int(np.ceil(max_y)) + 1

    if W <= 0 or H <= 0:
        return False

    def draw_mask(bboxes):
        mask = np.zeros((H, W), dtype=np.uint8)
        for box in bboxes:
            x1, y1, x2, y2 = box
            x1 = int(round(x1))
            y1 = int(round(y1))
            x2 = int(round(x2))
            y2 = int(round(y2))
            x1 = max(0, min(W, x1))
            x2 = max(0, min(W, x2))
            y1 = max(0, min(H, y1))
            y2 = max(0, min(H, y2))
            if x2 > x1 and y2 > y1:
                mask[y1:y2, x1:x2] = 1
        return mask

    mask1 = draw_mask(bboxes1)
    mask2 = draw_mask(bboxes2)

    diff = np.abs(mask1 - mask2)
    diff_ratio = np.sum(diff) / diff.size

    return diff_ratio <= diff_threshold


def get_model_output(model_name: str, input_image_path: str, **args):
    import json
    import shutil
    import pandas as pd

    results = pd.read_csv('benchmark/model_results.csv', sep=';')
    result = None

    try:
        # classification
        if model_name in ['MSCN', 'RemoteCLIP']:
            result = results[(results['model'] == model_name) & (results['file_path'] == input_image_path)]['result'].values[0]
            # result = json.loads(result)
        # detection
        elif model_name in ['Strip-R-CNN', 'SM3Det']:
            result = results[(results['model'] == model_name) & (results['file_path'] == input_image_path)]
            result = result[result['text_prompt'] == args['text_prompt']]['result'].values[0]
            result = json.loads(result)
        # visual grounding & counting
        elif model_name in ['RemoteSAM', 'InstructSAM']:
            result = results[(results['model'] == model_name) & (results['file_path'] == input_image_path)]
            result = result[result['text_prompt'] == args['text_prompt']]['result'].values[0]
            result = json.loads(result)
        # segmentation
        elif model_name == 'SAM2':
            result = results[(results['model'] == model_name) & (results['file_path'] == input_image_path)]
            bbox_csv = result['bbox'].values[0]
            bbox_input = json.loads(args['bbox']) if isinstance(args['bbox'], str) else args['bbox']
            if bboxes_equal(json.loads(bbox_csv), bbox_input):
                result = result['result'].values[0]
                (TEMP_DIR / args['output_path']).parent.mkdir(parents=True, exist_ok=True)
                shutil.copy(result, TEMP_DIR / args['output_path'])
                result = 'Result save at {}'.format(TEMP_DIR / args['output_path'])
            else:
                result = None
        # change detection
        elif model_name == 'ChangeOS':
            file_pair = '({}, {})'.format(input_image_path, args['post_image_path'])
            result = results[(results['model'] == model_name) & (results['file_path'] == file_pair)]['result'].values[0]
            (TEMP_DIR / args['output_path']).parent.mkdir(parents=True, exist_ok=True)
            shutil.copy(result, TEMP_DIR / args['output_path'])
            result = 'Result save at {}'.format(TEMP_DIR / args['output_path'])
        # change detection for building extraction
        elif model_name == 'ChangeOS_Building_Extraction':
            result = results[(results['model'] == 'ChangeOS') & (results['file_path'] == input_image_path)]['result'].values[0]
            (TEMP_DIR / args['output_path']).parent.mkdir(parents=True, exist_ok=True)
            shutil.copy(result, TEMP_DIR / args['output_path'])
            result = 'Result save at {}'.format(TEMP_DIR / args['output_path'])
    
        if result is None:
            return 'Failed to call model'
        else:
            return result
    except:
        return 'Failed to call model'



@mcp.tool(description="""
MSCN is a scene and land-use image classifier, effective for categories such as 
Airport, BareLand, BaseballField, Beach, Bridge, Center, Church, Commercial, 
DenseResidential, Desert, Farmland, Forest, Industrial, Meadow, MediumResidential, 
Mountain, Park, Parking, Playground, Pond, Port, RailwayStation, Resort, River, 
School, SparseResidential, Square, Stadium, StorageTanks, and Viaduct.
Parameters:
- input_image_path (str): Path to the input image.
Returns:
- np.ndarray: [model_name, image_path, predicted_class, confidence, top-5 predictions]
Example:
array([
  'MSCN',
  'benchmark/data/question189/J.jpg',
  'Resort',
  0.7052103281021118,
  [
    ('Resort', 0.7052103281021118),
    ('StorageTanks', 0.11459718644618988),
    ('Desert', 0.019159140065312386),
    ('Meadow', 0.013844668865203857),
    ('Beach', 0.013844599016010761)
  ]
], dtype=object)
""")
def MSCN(input_image_path):
    try:
        return get_model_output('MSCN', input_image_path)
    except Exception as e:

        return f'Error in MSCN: {str(e)}'


@mcp.tool(description="""
RemoteCLIP is a scene and land-use image classifier, specialized for categories such as 
Airport, Beach, Bridge, Commercial, Desert, Farmland, FootballField, Forest, Industrial, 
Meadow, Mountain, Park, Parking, Pond, Port, RailwayStation, Residential, River, and Viaduct.
Parameters:
- input_image_path (str): Path to the input image.
Returns:
- np.ndarray: [model_name, image_path, predicted_class, confidence, top-5 predictions]
Example:
array([
  'RemoteCLIP',
  'benchmark/data/question189/J.jpg',
  'Resort',
  0.7052103281021118,
  [
    ('Resort', 0.7052103281021118),
    ('StorageTanks', 0.11459718644618988),
    ('Desert', 0.019159140065312386),
    ('Meadow', 0.013844668865203857),
    ('Beach', 0.013844599016010761)
  ]
], dtype=object)
""")
def RemoteCLIP(input_image_path):
    try:
        return get_model_output('RemoteCLIP', input_image_path)
    except Exception as e:

        return f'Error in RemoteCLIP: {str(e)}'



@mcp.tool(description="""
Strip_R_CNN is a remote sensing object detection model with a strong focus on 
maritime and ship-related targets. Compared to SM3Det, it is particularly 
specialized in detecting and localizing different types of ships and naval vessels.
This model is highly effective at detecting the following categories:
- L3 ship
- L3 warcraft
- L3 merchant ship
- L3 aircraft carrier
- Arleigh Burke
- ContainerA
- Ticonderoga
- Perry
- Tarawa
- WhidbeyIsland
- CommanderA
- Austen
- Nimitz
- Sanantonio
- Container
- Car carrierB
- Enterprise
- Car carrierA
- Medical
Parameters:
- input_image_path (str): Path to the input image.
- text_prompt (str): Natural language description of the ship type to detect.
Returns:
- list[list[float]]: A list of bounding boxes, each represented as 
  [x_min, y_min, x_max, y_max].
Example:
Input:
  input_image_path = "benchmark/data/questionXXX/ship_example.png"
  text_prompt = "L3 aircraft carrier"
Output:
  [
    [120.5, 340.7, 480.2, 600.9],
    [700.3, 220.1, 950.6, 400.4]
  ]
""")
def Strip_R_CNN(input_image_path, text_prompt):
    try:
        return get_model_output('Strip-R-CNN', input_image_path, text_prompt=text_prompt)
    except Exception as e:

        return f'Error in Strip_R_CNN: {str(e)}'


@mcp.tool(description="""
SM3Det is a remote sensing object detection model. 
Given an input image and a natural language prompt specifying the target object 
(e.g., "plane", "ship", "storage tank"), it detects all instances of that object 
and returns their bounding boxes.
This model is particularly strong at detecting and localizing the following categories:
- plane
- ship
- storage tank
- baseball diamond
- tennis court
- basketball court
- ground track field
- harbor
- bridge
- large vehicle
- small vehicle
- helicopter
- roundabout
- soccer ball field
- swimming pool
Parameters:
- input_image_path (str): Path to the input image.
- text_prompt (str): Natural language description of the object to detect.
Returns:
- list[list[float]]: A list of bounding boxes, each represented as 
  [x_min, y_min, x_max, y_max].
Example:
Input:
  input_image_path = "benchmark/data/question235/P0173.png"
  text_prompt = "plane"
Output:
  [
    [491.08, 532.47, 562.03, 598.47],
    [548.89, 563.04, 636.54, 643.85],
    [57.80, 335.57, 191.65, 446.29],
    [401.37, 474.06, 509.87, 573.09],
    [344.69, 146.72, 464.14, 249.53],
    [736.10, 503.04, 809.28, 568.13],
    [680.84, 448.89, 760.03, 512.00],
    [588.72, 312.11, 666.23, 378.02],
    [537.49, 258.38, 610.34, 313.70]
  ]
""")
def SM3Det(input_image_path, text_prompt):
    try:
        return get_model_output('SM3Det', input_image_path, text_prompt=text_prompt)
    except Exception as e:

        return f'Error in SM3Det: {str(e)}'

@mcp.tool(description="""
RemoteSAM is a remote sensing visual grounding model. Given an input image and a text prompt 
describing a region of interest (e.g., "the football field located on the westernmost side"), 
it outputs the corresponding bounding box coordinates.
Parameters:
- input_image_path (str): Path to the input image.
- text_prompt (str): Natural language description of the target object/region.
Returns:
- list[int]: Bounding box [x_min, y_min, x_max, y_max]
Example:
Input:
  input_image_path = "benchmark/data/question226/478549_4934011_2048_32610_sport_soccer.jpg"
  text_prompt = "the football field located on the westernmost side"
Output:
  [0, 264, 127, 342]
""")
def RemoteSAM(input_image_path, text_prompt):
    try:
        return get_model_output('RemoteSAM', input_image_path, text_prompt=text_prompt)
    except Exception as e:

        return f'Error in RemoteSAM: {str(e)}'


@mcp.tool(description="""
InstructSAM is an instruction-guided counting model for remote sensing images. 
Given an input image and a natural language prompt specifying the target object 
(e.g., "storage tank", "football field"), it detects and counts the number of 
instances matching the description.
Parameters:
- input_image_path (str): Path to the input image.
- text_prompt (str): Natural language description of the object to count.
Returns:
- int: The number of objects in the image that match the text prompt.
Example:
Input:
  input_image_path = "benchmark/data/question231/B.jpg"
  text_prompt = "storage tank"
Output:
  8
""")
def InstructSAM(input_image_path, text_prompt):
    try:
        return get_model_output('InstructSAM', input_image_path, text_prompt=text_prompt)
    except Exception as e:

        return f'Error in InstructSAM: {str(e)}'


@mcp.tool(description="""
    Use SAM2 to segment the input image and return the bounding box.
    Parameters:
        input_image_path (str): Path to the input image.
        bboxes (list): Bounding box of the segmented object.
    Returns:
        str: Path to the segmented image.
""")
def SAM2(input_image_path, bboxes, output_path):
    try:
        return get_model_output('SAM2', input_image_path, bbox=bboxes, output_path=output_path)
    except Exception as e:

        return f'Error in SAM2: {str(e)}'


@mcp.tool(description="""
    Use ChangeOS to detect the change between two images and return the change mask.
    Can also be used to segment building by providing same image path in pre_image_path and post_image_path.
    Parameters:
        pre_image_path (str): Path to the pre-image.
        post_image_path (str): Path to the post-image.
        output_path (str): Path to the output change mask.
    Returns:
        str: Path to the segmented image.
""")
def ChangeOS(pre_image_path: str, post_image_path: str, output_path: str):
    try:
        if pre_image_path == post_image_path:
            return get_model_output('ChangeOS_Building_Extraction', pre_image_path, output_path=output_path)
        else:
            return get_model_output('ChangeOS', pre_image_path, post_image_path=post_image_path, output_path=output_path)
    except Exception as e:

        return f'Error in ChangeOS: {str(e)}'

if __name__ == "__main__":
    mcp.run(transport="http", host="127.0.0.1", port=args.port)