forked from Zardinality/TF_Deformable_Net
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
1 parent
cd3cdce
commit f7945b2
Showing
3 changed files
with
76 additions
and
64 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
104 changes: 52 additions & 52 deletions
104
lib/deform_psroi_pooling_layer/deform_psroi_pooling_op_grad.py
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,59 +1,59 @@ | ||
| from __future__ import absolute_import | ||
| import tensorflow as tf | ||
| from tensorflow.python.framework import ops | ||
| import deform_psroi_pooling_op | ||
| import pdb | ||
|
|
||
|
|
||
| # @tf.RegisterShape("DeformPSROIPool") | ||
| # def _deform_psroi_pool_shape(op): | ||
| # """Shape function for the DeformPSROIPool op. | ||
|
|
||
| # from __future__ import absolute_import | ||
| # import tensorflow as tf | ||
| # from tensorflow.python.framework import ops | ||
| # import deform_psroi_pooling_op | ||
| # import pdb | ||
|
|
||
|
|
||
| # # @tf.RegisterShape("DeformPSROIPool") | ||
| # # def _deform_psroi_pool_shape(op): | ||
| # # """Shape function for the DeformPSROIPool op. | ||
|
|
||
| # # """ | ||
| # # dims_data = op.inputs[0].get_shape().as_list() | ||
| # # channels = dims_data[3] | ||
| # # dims_rois = op.inputs[1].get_shape().as_list() | ||
| # # num_rois = dims_rois[0] | ||
| # # output_dim = op.get_attr('output_dim') | ||
| # # group_size = op.get_attr('group_size') | ||
| # # pooled_height = group_size | ||
| # # pooled_width = group_size | ||
|
|
||
| # # output_shape = tf.TensorShape([num_rois, pooled_height, pooled_width, output_dim]) | ||
| # # return [output_shape, output_shape] | ||
|
|
||
| # @ops.RegisterGradient("DeformPSROIPool") | ||
| # def _deform_psroi_pool_grad(op, grad, _): | ||
| # """The gradients for `Deform_PSROI_pool`. | ||
| # Args: | ||
| # op: The `roi_pool` `Operation` that we are differentiating, which we can use | ||
| # to find the inputs and outputs of the original op. | ||
| # grad: Gradient with respect to the output of the `roi_pool` op. | ||
| # Returns: | ||
| # Gradients with respect to the input of `zero_out`. | ||
| # """ | ||
| # dims_data = op.inputs[0].get_shape().as_list() | ||
| # channels = dims_data[3] | ||
| # dims_rois = op.inputs[1].get_shape().as_list() | ||
| # num_rois = dims_rois[0] | ||
| # output_dim = op.get_attr('output_dim') | ||
| # group_size = op.get_attr('group_size') | ||
| # pooled_height = group_size | ||
| # pooled_width = group_size | ||
|
|
||
| # output_shape = tf.TensorShape([num_rois, pooled_height, pooled_width, output_dim]) | ||
| # return [output_shape, output_shape] | ||
|
|
||
| @ops.RegisterGradient("DeformPSROIPool") | ||
| def _deform_psroi_pool_grad(op, grad, _): | ||
| """The gradients for `Deform_PSROI_pool`. | ||
| Args: | ||
| op: The `roi_pool` `Operation` that we are differentiating, which we can use | ||
| to find the inputs and outputs of the original op. | ||
| grad: Gradient with respect to the output of the `roi_pool` op. | ||
| Returns: | ||
| Gradients with respect to the input of `zero_out`. | ||
| """ | ||
|
|
||
|
|
||
| data = op.inputs[0] | ||
| rois = op.inputs[1] | ||
| trans = op.inputs[2] | ||
| mapping_channel = op.outputs[1] | ||
| spatial_scale = op.get_attr('spatial_scale') | ||
| output_dim = op.get_attr('output_dim') | ||
| group_size = op.get_attr('group_size') | ||
| pooled_size = op.get_attr('pooled_size') | ||
| part_size = op.get_attr('part_size') | ||
| sample_per_part = op.get_attr('sample_per_part') | ||
| trans_std = op.get_attr('trans_std') | ||
| no_trans = op.get_attr('no_trans') | ||
| # data = op.inputs[0] | ||
| # rois = op.inputs[1] | ||
| # trans = op.inputs[2] | ||
| # mapping_channel = op.outputs[1] | ||
| # spatial_scale = op.get_attr('spatial_scale') | ||
| # output_dim = op.get_attr('output_dim') | ||
| # group_size = op.get_attr('group_size') | ||
| # pooled_size = op.get_attr('pooled_size') | ||
| # part_size = op.get_attr('part_size') | ||
| # sample_per_part = op.get_attr('sample_per_part') | ||
| # trans_std = op.get_attr('trans_std') | ||
| # no_trans = op.get_attr('no_trans') | ||
|
|
||
|
|
||
|
|
||
| # compute gradient | ||
| #data_grad = psroi_pooling_op.psroi_pool_grad(data, rois, argmax, grad, pooled_height, pooled_width, spatial_scale) | ||
| data_grad, trans_grad = deform_psroi_pooling_op.deform_psroi_pool_grad(data, rois, trans, mapping_channel, grad, spatial_scale, | ||
| output_dim, group_size, pooled_size, part_size, sample_per_part, | ||
| trans_std, no_trans) | ||
| # rois_grad = tf.zeros(rois.shape) | ||
| return [data_grad, None, trans_grad] # List of one Tensor, since we have one input | ||
| # # compute gradient | ||
| # #data_grad = psroi_pooling_op.psroi_pool_grad(data, rois, argmax, grad, pooled_height, pooled_width, spatial_scale) | ||
| # data_grad, trans_grad = deform_psroi_pooling_op.deform_psroi_pool_grad(data, rois, trans, mapping_channel, grad, spatial_scale, | ||
| # output_dim, group_size, pooled_size, part_size, sample_per_part, | ||
| # trans_std, no_trans) | ||
| # # rois_grad = tf.zeros(rois.shape) | ||
| # return [data_grad, None, trans_grad] # List of one Tensor, since we have one input | ||
|
|
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters