Skip to content

vllm.multimodal.image

ImageEmbeddingMediaIO

Bases: MediaIO[Tensor]

Source code in vllm/multimodal/image.py 
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
class ImageEmbeddingMediaIO(MediaIO[torch.Tensor]):
    def __init__(self) -> None:
        super().__init__()

    def load_bytes(self, data: bytes) -> torch.Tensor:
        buffer = BytesIO(data)
        return torch.load(buffer, weights_only=True)

    def load_base64(self, media_type: str, data: str) -> torch.Tensor:
        return self.load_bytes(pybase64.b64decode(data, validate=True))

    def load_file(self, filepath: Path) -> torch.Tensor:
        return torch.load(filepath, weights_only=True)

    def encode_base64(self, media: torch.Tensor) -> str:
        return pybase64.b64encode(media.numpy()).decode("utf-8")

__init__ 

__init__() -> None
Source code in vllm/multimodal/image.py 
116
117
def __init__(self) -> None:
    super().__init__()

encode_base64 

encode_base64(media: Tensor) -> str
Source code in vllm/multimodal/image.py 
129
130
def encode_base64(self, media: torch.Tensor) -> str:
    return pybase64.b64encode(media.numpy()).decode("utf-8")

load_base64 

load_base64(media_type: str, data: str) -> Tensor
Source code in vllm/multimodal/image.py 
123
124
def load_base64(self, media_type: str, data: str) -> torch.Tensor:
    return self.load_bytes(pybase64.b64decode(data, validate=True))

load_bytes 

load_bytes(data: bytes) -> Tensor
Source code in vllm/multimodal/image.py 
119
120
121
def load_bytes(self, data: bytes) -> torch.Tensor:
    buffer = BytesIO(data)
    return torch.load(buffer, weights_only=True)

load_file 

load_file(filepath: Path) -> Tensor
Source code in vllm/multimodal/image.py 
126
127
def load_file(self, filepath: Path) -> torch.Tensor:
    return torch.load(filepath, weights_only=True)

ImageMediaIO

Bases: MediaIO[Image]

Source code in vllm/multimodal/image.py 
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
class ImageMediaIO(MediaIO[Image.Image]):
    def __init__(self, image_mode: str = "RGB", **kwargs) -> None:
        super().__init__()

        self.image_mode = image_mode
        # `kwargs` contains custom arguments from
        # --media-io-kwargs for this modality.
        # They can be passed to the underlying
        # media loaders (e.g. custom implementations)
        # for flexible control.
        self.kwargs = kwargs

        # Extract RGBA background color from kwargs if provided
        # Default to white background for backward compatibility
        rgba_bg = kwargs.get("rgba_background_color", (255, 255, 255))
        # Convert list to tuple for consistency
        if isinstance(rgba_bg, list):
            rgba_bg = tuple(rgba_bg)

        # Validate rgba_background_color format
        if not (
            isinstance(rgba_bg, tuple)
            and len(rgba_bg) == 3
            and all(isinstance(c, int) and 0 <= c <= 255 for c in rgba_bg)
        ):
            raise ValueError(
                "rgba_background_color must be a list or tuple of 3 integers "
                "in the range [0, 255]."
            )
        self.rgba_background_color = rgba_bg

    def _convert_image_mode(self, image: Image.Image) -> Image.Image:
        """Convert image mode with custom background color."""
        if image.mode == self.image_mode:
            return image
        elif image.mode == "RGBA" and self.image_mode == "RGB":
            return rgba_to_rgb(image, self.rgba_background_color)
        else:
            return convert_image_mode(image, self.image_mode)

    def load_bytes(self, data: bytes) -> Image.Image:
        image = Image.open(BytesIO(data))
        image.load()
        return self._convert_image_mode(image)

    def load_base64(self, media_type: str, data: str) -> Image.Image:
        return self.load_bytes(pybase64.b64decode(data, validate=True))

    def load_file(self, filepath: Path) -> Image.Image:
        image = Image.open(filepath)
        image.load()
        return self._convert_image_mode(image)

    def encode_base64(
        self,
        media: Image.Image,
        *,
        image_format: str = "JPEG",
    ) -> str:
        image = media

        with BytesIO() as buffer:
            image = self._convert_image_mode(image)
            image.save(buffer, image_format)
            data = buffer.getvalue()

        return pybase64.b64encode(data).decode("utf-8")

image_mode instance-attribute 

image_mode = image_mode

kwargs instance-attribute 

kwargs = kwargs

rgba_background_color instance-attribute 

rgba_background_color = rgba_bg

__init__ 

__init__(image_mode: str = 'RGB', **kwargs) -> None
Source code in vllm/multimodal/image.py 
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
def __init__(self, image_mode: str = "RGB", **kwargs) -> None:
    super().__init__()

    self.image_mode = image_mode
    # `kwargs` contains custom arguments from
    # --media-io-kwargs for this modality.
    # They can be passed to the underlying
    # media loaders (e.g. custom implementations)
    # for flexible control.
    self.kwargs = kwargs

    # Extract RGBA background color from kwargs if provided
    # Default to white background for backward compatibility
    rgba_bg = kwargs.get("rgba_background_color", (255, 255, 255))
    # Convert list to tuple for consistency
    if isinstance(rgba_bg, list):
        rgba_bg = tuple(rgba_bg)

    # Validate rgba_background_color format
    if not (
        isinstance(rgba_bg, tuple)
        and len(rgba_bg) == 3
        and all(isinstance(c, int) and 0 <= c <= 255 for c in rgba_bg)
    ):
        raise ValueError(
            "rgba_background_color must be a list or tuple of 3 integers "
            "in the range [0, 255]."
        )
    self.rgba_background_color = rgba_bg

_convert_image_mode 

_convert_image_mode(image: Image) -> Image

Convert image mode with custom background color.

Source code in vllm/multimodal/image.py 
77
78
79
80
81
82
83
84
def _convert_image_mode(self, image: Image.Image) -> Image.Image:
    """Convert image mode with custom background color."""
    if image.mode == self.image_mode:
        return image
    elif image.mode == "RGBA" and self.image_mode == "RGB":
        return rgba_to_rgb(image, self.rgba_background_color)
    else:
        return convert_image_mode(image, self.image_mode)

encode_base64 

encode_base64(
    media: Image, *, image_format: str = "JPEG"
) -> str
Source code in vllm/multimodal/image.py 
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
def encode_base64(
    self,
    media: Image.Image,
    *,
    image_format: str = "JPEG",
) -> str:
    image = media

    with BytesIO() as buffer:
        image = self._convert_image_mode(image)
        image.save(buffer, image_format)
        data = buffer.getvalue()

    return pybase64.b64encode(data).decode("utf-8")

load_base64 

load_base64(media_type: str, data: str) -> Image
Source code in vllm/multimodal/image.py 
91
92
def load_base64(self, media_type: str, data: str) -> Image.Image:
    return self.load_bytes(pybase64.b64decode(data, validate=True))

load_bytes 

load_bytes(data: bytes) -> Image
Source code in vllm/multimodal/image.py 
86
87
88
89
def load_bytes(self, data: bytes) -> Image.Image:
    image = Image.open(BytesIO(data))
    image.load()
    return self._convert_image_mode(image)

load_file 

load_file(filepath: Path) -> Image
Source code in vllm/multimodal/image.py 
94
95
96
97
def load_file(self, filepath: Path) -> Image.Image:
    image = Image.open(filepath)
    image.load()
    return self._convert_image_mode(image)

convert_image_mode 

convert_image_mode(image: Image, to_mode: str)
Source code in vllm/multimodal/image.py 
37
38
39
40
41
42
43
def convert_image_mode(image: Image.Image, to_mode: str):
    if image.mode == to_mode:
        return image
    elif image.mode == "RGBA" and to_mode == "RGB":
        return rgba_to_rgb(image)
    else:
        return image.convert(to_mode)

rescale_image_size 

rescale_image_size(
    image: Image, size_factor: float, transpose: int = -1
) -> Image

Rescale the dimensions of an image by a constant factor.

Source code in vllm/multimodal/image.py 
14
15
16
17
18
19
20
21
22
23
def rescale_image_size(
    image: Image.Image, size_factor: float, transpose: int = -1
) -> Image.Image:
    """Rescale the dimensions of an image by a constant factor."""
    new_width = int(image.width * size_factor)
    new_height = int(image.height * size_factor)
    image = image.resize((new_width, new_height))
    if transpose >= 0:
        image = image.transpose(Image.Transpose(transpose))
    return image

rgba_to_rgb 

rgba_to_rgb(
    image: Image,
    background_color: tuple[int, int, int] | list[int] = (
        255,
        255,
        255,
    ),
) -> Image

Convert an RGBA image to RGB with filled background color.

Source code in vllm/multimodal/image.py 
26
27
28
29
30
31
32
33
34
def rgba_to_rgb(
    image: Image.Image,
    background_color: tuple[int, int, int] | list[int] = (255, 255, 255),
) -> Image.Image:
    """Convert an RGBA image to RGB with filled background color."""
    assert image.mode == "RGBA"
    converted = Image.new("RGB", image.size, background_color)
    converted.paste(image, mask=image.split()[3])  # 3 is the alpha channel
    return converted