deeplabv3_mobilenet_v3_large(*, weights: Optional[torchvision.models.segmentation.deeplabv3.DeepLabV3_MobileNet_V3_Large_Weights] = None, progress: bool = True, num_classes: Optional[int] = None, aux_loss: Optional[bool] = None, weights_backbone: Optional[torchvision.models.mobilenetv3.MobileNet_V3_Large_Weights] = MobileNet_V3_Large_Weights.IMAGENET1K_V1, **kwargs: Any) → torchvision.models.segmentation.deeplabv3.DeepLabV3¶
Constructs a DeepLabV3 model with a MobileNetV3-Large backbone.
DeepLabV3_MobileNet_V3_Large_Weights, optional) – The pretrained weights to use. See
DeepLabV3_MobileNet_V3_Large_Weightsbelow for more details, and possible values. By default, no pre-trained weights are used.
num_classes (int, optional) – number of output classes of the model (including the background)
aux_loss (bool, optional) – If True, it uses an auxiliary loss
MobileNet_V3_Large_Weights, optional) – The pretrained weights for the backbone
**kwargs – unused
The model builder above accepts the following values as the
DeepLabV3_MobileNet_V3_Large_Weights.DEFAULTis equivalent to
DeepLabV3_MobileNet_V3_Large_Weights.COCO_WITH_VOC_LABELS_V1. You can also use strings, e.g.
These weights were trained on a subset of COCO, using only the 20 categories that are present in the Pascal VOC dataset. Also available as
miou (on COCO-val2017-VOC-labels)
pixel_acc (on COCO-val2017-VOC-labels)
__background__, aeroplane, bicycle, … (18 omitted)
The inference transforms are available at
DeepLabV3_MobileNet_V3_Large_Weights.COCO_WITH_VOC_LABELS_V1.transformsand perform the following preprocessing operations: Accepts
(B, C, H, W)and single
(C, H, W)image
torch.Tensorobjects. The images are resized to
interpolation=InterpolationMode.BILINEAR. Finally the values are first rescaled to
[0.0, 1.0]and then normalized using
mean=[0.485, 0.456, 0.406]and
std=[0.229, 0.224, 0.225].