vis4d.model.detect.mask_rcnn¶

Mask RCNN model implementation and runtime.

Classes

`MaskDetectionOut`(boxes, masks)	Mask detection output.
`MaskRCNN`(num_classes[, basemodel, ...])	Mask RCNN model.
`MaskRCNNOut`(boxes, masks)	Mask RCNN output.

class MaskDetectionOut(boxes: DetOut, masks: MaskOut)[source]¶

Mask detection output.

Create new instance of MaskDetectionOut(boxes, masks)

boxes: DetOut¶: Alias for field number 0

masks: MaskOut¶: Alias for field number 1

class MaskRCNN(num_classes, basemodel=None, faster_rcnn_head=None, mask_head=None, rcnn_box_decoder=None, no_overlap=False, weights=None)[source]¶

Mask RCNN model.

Parameters:

num_classes (int) – Number of classes.
basemodel (BaseModel, optional) – Base model network. Defaults to None. If None, will use ResNet50.
faster_rcnn_head (FasterRCNNHead, optional) – Faster RCNN head. Defaults to None. if None, will use default FasterRCNNHead.
mask_head (MaskRCNNHead, optional) – Mask RCNN head. Defaults to None. if None, will use default MaskRCNNHead.
rcnn_box_decoder (DeltaXYWHBBoxDecoder, optional) – Decoder for RCNN bounding boxes. Defaults to None.
no_overlap (bool, optional) – Whether to remove overlapping pixels between masks. Defaults to False.
weights (None | str, optional) – Weights to load for model. If set to “mmdet”, will load MMDetection pre-trained weights. Defaults to None.

Creates an instance of the class.

forward(images, input_hw, boxes2d=None, boxes2d_classes=None, original_hw=None)[source]¶

Forward pass.

Parameters:

images (torch.Tensor) – Input images.
input_hw (list[tuple[int, int]]) – Input image resolutions.
boxes2d (None | list[torch.Tensor], optional) – Bounding box labels. Required for training. Defaults to None.
boxes2d_classes (None | list[torch.Tensor], optional) – Class labels. Required for training. Defaults to None.
original_hw (None | list[tuple[int, int]], optional) – Original image resolutions (before padding and resizing). Required for testing. Defaults to None.

Returns:

Either raw model: outputs (for training) or predicted outputs (for testing).

Return type:

MaskRCNNOut | MaskDetectionOut

forward_test(images, images_hw, original_hw)[source]¶

Forward testing stage.

Parameters:

images (torch.Tensor) – Input images.
images_hw (list[tuple[int, int]]) – Input image resolutions.
original_hw (list[tuple[int, int]]) – Original image resolutions (before padding and resizing).

Returns:

Predicted outputs.

Return type:

MaskDetectionOut

forward_train(images, images_hw, target_boxes, target_classes)[source]¶

Forward training stage.

Parameters:

images (torch.Tensor) – Input images.
images_hw (list[tuple[int, int]]) – Input image resolutions.
target_boxes (list[torch.Tensor]) – Bounding box labels. Required for training. Defaults to None.
target_classes (list[torch.Tensor]) – Class labels. Required for training. Defaults to None.

Returns:

Raw model outputs.

Return type:

MaskRCNNOut

class MaskRCNNOut(boxes: FRCNNOut, masks: MaskRCNNHeadOut)[source]¶

Mask RCNN output.

Create new instance of MaskRCNNOut(boxes, masks)

boxes: FRCNNOut¶: Alias for field number 0

masks: MaskRCNNHeadOut¶: Alias for field number 1