tfm.vision.target_gather.TargetGather

Targer gather for dense object detector.

Methods

__call__(
    labels, match_indices, mask=None, mask_val=0.0
)

Labels anchors with ground truth inputs.

B: batch_size N: number of groundtruth boxes.

Args
`labels`	An integer tensor with shape [N, dims] or [B, N, ...] representing groundtruth labels.
`match_indices`	An integer tensor with shape [M] or [B, M] representing match label index.
`mask`	An boolean tensor with shape [M, dims] or [B, M,...] representing match labels.
`mask_val`	An integer to fill in for mask.

Returns
`target`	An integer Tensor with shape [M] or [B, M]

Raises
`ValueError`	If `labels` is higher than rank 3.

Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates. Some content is licensed under the numpy license.

Last updated 2024-02-02 UTC.