Questions about api about open-cd HOT 4 CLOSED

Sorry for the delay in replying to you, you can try the following code.

import mmcv
import torch
from mmcv.parallel import collate, scatter

from mmseg.apis import init_segmentor
from mmseg.datasets.pipelines import Compose
from opencd.models import *


class LoadImage:
    """A simple pipeline to load image."""

    def __call__(self, results):
        """Call function to load images into results.

        Args:
            results (dict): A result dict contains the file name
                of the image to be read.

        Returns:
            dict: ``results`` will be returned containing loaded image.
        """

        results['filename'] = None
        results['ori_filename'] = None
        results['img'] = img
        results['img_shape'] = img[0].shape
        results['ori_shape'] = img[0].shape
        return results


def inference_segmentor(model, imgs):
    """Inference image(s) with the segmentor.

    Args:
        model (nn.Module): The loaded segmentor.
        imgs (str/ndarray or list[str/ndarray]): Either image files or loaded
            images.

    Returns:
        (list[Tensor]): The segmentation result.
    """
    cfg = model.cfg
    device = next(model.parameters()).device  # model device
    # build the data pipeline
    test_pipeline = [LoadImage()] + cfg.data.test.pipeline[1:]
    test_pipeline = Compose(test_pipeline)
    # prepare data
    data = []
    imgs = imgs if isinstance(imgs[0], list) else [imgs]
    for img in imgs:
        img_data = dict(img=img)
        img_data = test_pipeline(img_data)
        data.append(img_data)
    data = collate(data, samples_per_gpu=len(imgs))
    if next(model.parameters()).is_cuda:
        # scatter to specified GPU
        data = scatter(data, [device])[0]
    else:
        data['img_metas'] = [i.data[0] for i in data['img_metas']]

    # forward the model
    with torch.no_grad():
        result = model(return_loss=False, rescale=True, **data)
    return result


config_file = 'configs/changer/changer_ex_r18_512x512_40k_levircd.py'
checkpoint_file = 'weights/ChangerEx_r18-512x512_40k_levircd_20221223_120511.pth'

# build the model from a config file and a checkpoint file
model = init_segmentor(config_file, checkpoint_file, device='cuda:0')

# test a single image and show the results
img1 = mmcv.imread('data/LEVIR-CD/test/A/test_1.png')
img2 = mmcv.imread('data/LEVIR-CD/test/B/test_1.png')
img = [img1, img2]

result = inference_segmentor(model, img)

Thanks for your replying! I have already completed the api by myself, you can close this issue~.

The example code provided above uses an outdated function, so it is difficult to use it as is. Can you share the API you implemented?

from open-cd.

Sorry for the delay in replying to you, you can try the following code.

import mmcv
import torch
from mmcv.parallel import collate, scatter

from mmseg.apis import init_segmentor
from mmseg.datasets.pipelines import Compose
from opencd.models import *


class LoadImage:
    """A simple pipeline to load image."""

    def __call__(self, results):
        """Call function to load images into results.

        Args:
            results (dict): A result dict contains the file name
                of the image to be read.

        Returns:
            dict: ``results`` will be returned containing loaded image.
        """

        results['filename'] = None
        results['ori_filename'] = None
        results['img'] = img
        results['img_shape'] = img[0].shape
        results['ori_shape'] = img[0].shape
        return results


def inference_segmentor(model, imgs):
    """Inference image(s) with the segmentor.

    Args:
        model (nn.Module): The loaded segmentor.
        imgs (str/ndarray or list[str/ndarray]): Either image files or loaded
            images.

    Returns:
        (list[Tensor]): The segmentation result.
    """
    cfg = model.cfg
    device = next(model.parameters()).device  # model device
    # build the data pipeline
    test_pipeline = [LoadImage()] + cfg.data.test.pipeline[1:]
    test_pipeline = Compose(test_pipeline)
    # prepare data
    data = []
    imgs = imgs if isinstance(imgs[0], list) else [imgs]
    for img in imgs:
        img_data = dict(img=img)
        img_data = test_pipeline(img_data)
        data.append(img_data)
    data = collate(data, samples_per_gpu=len(imgs))
    if next(model.parameters()).is_cuda:
        # scatter to specified GPU
        data = scatter(data, [device])[0]
    else:
        data['img_metas'] = [i.data[0] for i in data['img_metas']]

    # forward the model
    with torch.no_grad():
        result = model(return_loss=False, rescale=True, **data)
    return result


config_file = 'configs/changer/changer_ex_r18_512x512_40k_levircd.py'
checkpoint_file = 'weights/ChangerEx_r18-512x512_40k_levircd_20221223_120511.pth'

# build the model from a config file and a checkpoint file
model = init_segmentor(config_file, checkpoint_file, device='cuda:0')

# test a single image and show the results
img1 = mmcv.imread('data/LEVIR-CD/test/A/test_1.png')
img2 = mmcv.imread('data/LEVIR-CD/test/B/test_1.png')
img = [img1, img2]

result = inference_segmentor(model, img)

from open-cd.

Sorry for the delay in replying to you, you can try the following code.

import mmcv
import torch
from mmcv.parallel import collate, scatter

from mmseg.apis import init_segmentor
from mmseg.datasets.pipelines import Compose
from opencd.models import *


class LoadImage:
    """A simple pipeline to load image."""

    def __call__(self, results):
        """Call function to load images into results.

        Args:
            results (dict): A result dict contains the file name
                of the image to be read.

        Returns:
            dict: ``results`` will be returned containing loaded image.
        """

        results['filename'] = None
        results['ori_filename'] = None
        results['img'] = img
        results['img_shape'] = img[0].shape
        results['ori_shape'] = img[0].shape
        return results


def inference_segmentor(model, imgs):
    """Inference image(s) with the segmentor.

    Args:
        model (nn.Module): The loaded segmentor.
        imgs (str/ndarray or list[str/ndarray]): Either image files or loaded
            images.

    Returns:
        (list[Tensor]): The segmentation result.
    """
    cfg = model.cfg
    device = next(model.parameters()).device  # model device
    # build the data pipeline
    test_pipeline = [LoadImage()] + cfg.data.test.pipeline[1:]
    test_pipeline = Compose(test_pipeline)
    # prepare data
    data = []
    imgs = imgs if isinstance(imgs[0], list) else [imgs]
    for img in imgs:
        img_data = dict(img=img)
        img_data = test_pipeline(img_data)
        data.append(img_data)
    data = collate(data, samples_per_gpu=len(imgs))
    if next(model.parameters()).is_cuda:
        # scatter to specified GPU
        data = scatter(data, [device])[0]
    else:
        data['img_metas'] = [i.data[0] for i in data['img_metas']]

    # forward the model
    with torch.no_grad():
        result = model(return_loss=False, rescale=True, **data)
    return result


config_file = 'configs/changer/changer_ex_r18_512x512_40k_levircd.py'
checkpoint_file = 'weights/ChangerEx_r18-512x512_40k_levircd_20221223_120511.pth'

# build the model from a config file and a checkpoint file
model = init_segmentor(config_file, checkpoint_file, device='cuda:0')

# test a single image and show the results
img1 = mmcv.imread('data/LEVIR-CD/test/A/test_1.png')
img2 = mmcv.imread('data/LEVIR-CD/test/B/test_1.png')
img = [img1, img2]

result = inference_segmentor(model, img)

Thanks for your replying!
I have already completed the api by myself, you can close this issue~.

from open-cd.

Hi~, Open-CD supports inference api now 😃 ! see here

from open-cd.