AttributeError: Can't pickle local object 'TrainAugmentation.__init__.<locals>.<lambda>'

[Zx117x]

The following error occurred during training(python train_ssd.py)
AttributeError: Can't pickle local object 'TrainAugmentation.init..'
EOFError: Ran out of input

[littlerain2310]

can u fix it ?? i have the same issue

[wangwangww]

I also have this quesion.

[littlerain2310]

i read a comment saying that this program need the Os to be Linux(It doesnt work on Window)

[EnricoBeltramo]

add to command line option: --num_workers=0
similar issue:
pyg-team/pytorch_geometric#366

[laolihaile]

I also have this quesion.

[kueblert]

Replacing the lambda function in TrainAugmentation with the following function worked for me:

class ScaleByStd:
def init(self, std):
self.std = std

def __call__(self, image, boxes=None, labels=None):
    return image / self.std, boxes, labels

[jamespreed]

Full instructions based on @kueblert 's answer:

In the file vision\ssd\data_preprocessing.py insert the following lines after the import statement:

class ScaleByStd:
    def __init__(self, std: float):
        self.std = std

    def __call__(self, img, boxes=None, labels=None):
        return (img / self.std, boxes, labels)

Next modify the definitions of the TrainAugmentation, TestTransform and PredictionTransform classes by replacing the lines:

lambda img, boxes=None, labels=None: (img / std, boxes, labels),

with

ScaleByStd(std),

The top of the file should look like this:

from ..transforms.transforms import *


class ScaleByStd:
    def __init__(self, std: float):
        self.std = std

    def __call__(self, img, boxes=None, labels=None):
        return (img / self.std, boxes, labels)


class TrainAugmentation:
    def __init__(self, size, mean=0, std=1.0):
        """
        Args:
            size: the size the of final image.
            mean: mean pixel value per channel.
        """
        self.mean = mean
        self.size = size
        self.augment = Compose([
            ConvertFromInts(),
            PhotometricDistort(),
            Expand(self.mean),
            RandomSampleCrop(),
            RandomMirror(),
            ToPercentCoords(),
            Resize(self.size),
            SubtractMeans(self.mean),
            ScaleByStd(std),
            ToTensor(),
        ])

    def __call__(self, img, boxes, labels):
        """

        Args:
            img: the output of cv.imread in RGB layout.
            boxes: boundding boxes in the form of (x1, y1, x2, y2).
            labels: labels of boxes.
        """
        return self.augment(img, boxes, labels)


class TestTransform:
    def __init__(self, size, mean=0.0, std=1.0):
        self.transform = Compose([
            ToPercentCoords(),
            Resize(size),
            SubtractMeans(mean),
            ScaleByStd(std),
            ToTensor(),
        ])

    def __call__(self, image, boxes, labels):
        return self.transform(image, boxes, labels)


class PredictionTransform:
    def __init__(self, size, mean=0.0, std=1.0):
        self.transform = Compose([
            Resize(size),
            SubtractMeans(mean),
            ScaleByStd(std),
            ToTensor()
        ])

    def __call__(self, image):
        image, _, _ = self.transform(image)
        return image

[rsamvelyan]

Full instructions based on @kueblert 's answer:

In the file vision\ssd\data_preprocessing.py insert the following lines after the import statement:

class ScaleByStd:
    def __init__(self, std: float):
        self.std = std

    def __call__(self, img, boxes=None, labels=None):
        return (img / self.std, boxes, labels)

Next modify the definitions of the TrainAugmentation, TestTransform and PredictionTransform classes by replacing the lines:

lambda img, boxes=None, labels=None: (img / std, boxes, labels),

with

ScaleByStd(std),

The top of the file should look like this:

from ..transforms.transforms import *


class ScaleByStd:
    def __init__(self, std: float):
        self.std = std

    def __call__(self, img, boxes=None, labels=None):
        return (img / self.std, boxes, labels)


class TrainAugmentation:
    def __init__(self, size, mean=0, std=1.0):
        """
        Args:
            size: the size the of final image.
            mean: mean pixel value per channel.
        """
        self.mean = mean
        self.size = size
        self.augment = Compose([
            ConvertFromInts(),
            PhotometricDistort(),
            Expand(self.mean),
            RandomSampleCrop(),
            RandomMirror(),
            ToPercentCoords(),
            Resize(self.size),
            SubtractMeans(self.mean),
            ScaleByStd(std),
            ToTensor(),
        ])

    def __call__(self, img, boxes, labels):
        """

        Args:
            img: the output of cv.imread in RGB layout.
            boxes: boundding boxes in the form of (x1, y1, x2, y2).
            labels: labels of boxes.
        """
        return self.augment(img, boxes, labels)


class TestTransform:
    def __init__(self, size, mean=0.0, std=1.0):
        self.transform = Compose([
            ToPercentCoords(),
            Resize(size),
            SubtractMeans(mean),
            ScaleByStd(std),
            ToTensor(),
        ])

    def __call__(self, image, boxes, labels):
        return self.transform(image, boxes, labels)


class PredictionTransform:
    def __init__(self, size, mean=0.0, std=1.0):
        self.transform = Compose([
            Resize(size),
            SubtractMeans(mean),
            ScaleByStd(std),
            ToTensor()
        ])

    def __call__(self, image):
        image, _, _ = self.transform(image)
        return image

It worked for me!
I'm getting a warning though on line 247 of transforms.py:

...you must specify data type=object when creating the ndarray...

[Zx117x]

您好，您的邮件我已收到，我将尽快查看给予回复

[Neltherion]

@rsamvelyan Thanks. Any idea why your code change made the error go away?