utils/torch_utils/ops/grid_sample_gradfix.py

# SPDX-FileCopyrightText: Copyright (c) 2021-2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: LicenseRef-NvidiaProprietary
#
# NVIDIA CORPORATION, its affiliates and licensors retain all intellectual
# property and proprietary rights in and to this material, related
# documentation and any modifications thereto. Any use, reproduction,
# disclosure or distribution of this material and related documentation
# without an express license agreement from NVIDIA CORPORATION or
# its affiliates is strictly prohibited.
"""Custom replacement for `torch.nn.functional.grid_sample` that
supports arbitrarily high order gradients between the input and output.
Only works on 2D images and assumes
`mode='bilinear'`, `padding_mode='zeros'`, `align_corners=False`."""

import torch

# pylint: disable=redefined-builtin
# pylint: disable=arguments-differ
# pylint: disable=protected-access

#----------------------------------------------------------------------------

enabled = False  # Enable the custom op by setting this to true.

#----------------------------------------------------------------------------


def grid_sample(input, grid):
    if _should_use_custom_op():
        return _GridSample2dForward.apply(input, grid)
    return torch.nn.functional.grid_sample(input=input,
                                           grid=grid,
                                           mode='bilinear',
                                           padding_mode='zeros',
                                           align_corners=False)


#----------------------------------------------------------------------------


def _should_use_custom_op():
    return enabled


#----------------------------------------------------------------------------


class _GridSample2dForward(torch.autograd.Function):
    @staticmethod
    def forward(ctx, input, grid):
        assert input.ndim == 4
        assert grid.ndim == 4
        output = torch.nn.functional.grid_sample(input=input,
                                                 grid=grid,
                                                 mode='bilinear',
                                                 padding_mode='zeros',
                                                 align_corners=False)
        ctx.save_for_backward(input, grid)
        return output

    @staticmethod
    def backward(ctx, grad_output):
        input, grid = ctx.saved_tensors
        grad_input, grad_grid = _GridSample2dBackward.apply(
            grad_output, input, grid)
        return grad_input, grad_grid


#----------------------------------------------------------------------------


class _GridSample2dBackward(torch.autograd.Function):
    @staticmethod
    def forward(ctx, grad_output, input, grid):
        op = torch._C._jit_get_operation('aten::grid_sampler_2d_backward')
        grad_input, grad_grid = op(grad_output, input, grid, 0, 0, False)
        ctx.save_for_backward(grid)
        return grad_input, grad_grid

    @staticmethod
    def backward(ctx, grad2_grad_input, grad2_grad_grid):
        _ = grad2_grad_grid  # unused
        grid, = ctx.saved_tensors
        grad2_grad_output = None
        grad2_input = None
        grad2_grid = None

        if ctx.needs_input_grad[0]:
            grad2_grad_output = _GridSample2dForward.apply(
                grad2_grad_input, grid)

        assert not ctx.needs_input_grad[2]
        return grad2_grad_output, grad2_input, grad2_grid


#----------------------------------------------------------------------------