def _import_prroi_pooling():
global _prroi_pooling
if _prroi_pooling is None:
try:
from os.path import join as pjoin, dirname
from torch.utils.cpp_extension import load as load_extension
root_dir = pjoin(dirname(__file__), 'src')
_prroi_pooling = load_extension('_prroi_pooling', [
pjoin(root_dir, 'prroi_pooling_gpu.c'),
pjoin(root_dir, 'prroi_pooling_gpu_impl.cu')
],
verbose=True)
except ImportError:
raise ImportError('Can not compile Precise RoI Pooling library.')
return _prroi_pooling
import torch
from torch.autograd import Function
from torch.nn.modules.utils import _pair
import cffi
# from .._ext import depthconv
import torch.autograd as ag
try:
from os.path import join as pjoin, dirname
from torch.utils.cpp_extension import load as load_extension
root_dir = pjoin(dirname(__file__), '../src_pytorch13')
depthconv = load_extension('_depthconv', [
pjoin(root_dir, 'depthconv_cuda_redo.c'),
pjoin(root_dir, 'depthconv_cuda_kernel.cu')
],
verbose=True)
except ImportError:
raise ImportError('Can not compile depth-aware cnn library.')
__all__ = ['depth_conv']
def depth_conv(input, depth, weight, bias, stride=1, padding=0, dilation=1):
if input is not None and input.dim() != 4:
raise ValueError(
"Expected 4D tensor as input, got {}D tensor instead.".format(
input.dim()))
f = DepthconvFunction(_pair(stride), _pair(padding), _pair(dilation))
import torch
try:
from os.path import join as pjoin, dirname
from torch.utils.cpp_extension import load as load_extension
root_dir = pjoin(dirname(__file__), 'src')
_psroi_pooling = load_extension(
'_psroi_pooling', [
pjoin(root_dir, 'psroi_pooling_cuda.c'),
pjoin(root_dir, 'cuda/psroi_pooling_kernel.cu')
], verbose=True,
)
except ImportError:
raise ImportError('Can not compile Position Sensitive RoI Pooling library.')
class PsRoIPool2DFunction(torch.autograd.Function):
def __init__(self, pooled_height: int, pooled_width: int, spatial_scale: float,
group_size: int, output_dim: int):
self.pooled_width = int(pooled_width)
self.pooled_height = int(pooled_height)
self.spatial_scale = float(spatial_scale)
self.group_size = int(group_size)
self.output_dim = int(output_dim)
self.output = None
self.mappingchannel = None
self.rois = None
import torch
try:
from os.path import join as pjoin, dirname
from torch.utils.cpp_extension import load as load_extension
root_dir = pjoin(dirname(__file__), 'src')
_prroi_pooling = load_extension(
'_prroi_pooling', [
pjoin(root_dir, 'prroi_pooling_gpu.c'),
pjoin(root_dir, 'prroi_pooling_gpu_impl.cu')
],
)
except ImportError:
raise ImportError('Can not compile Precise RoI Pooling library.')
__all__ = ['prroi_pool2d']
class PrRoIPool2DFunction(torch.autograd.Function):
@staticmethod
def forward(ctx, features, rois, pooled_height, pooled_width, spatial_scale):
assert isinstance(features, torch.cuda.FloatTensor) or isinstance(rois, torch.cuda.FloatTensor), \
'Precise RoI Pooling only takes float input, got {} for features and {} for rois.'.format(features.type(), rois.type())
pooled_height = int(pooled_height)
pooled_width = int(pooled_width)
spatial_scale = float(spatial_scale)
features = features.contiguous()
rois = rois.contiguous()
# Date : 07/13/2018
#
# This file is part of PreciseRoIPooling.
# Distributed under terms of the MIT license.
# Copyright (c) 2017 Megvii Technology Limited.
import torch
import torch.autograd as ag
try:
from os.path import join as pjoin, dirname
from torch.utils.cpp_extension import load as load_extension
root_dir = pjoin(dirname(__file__), 'src')
_prroi_pooling = load_extension(
'_prroi_pooling',
[pjoin(root_dir, 'prroi_pooling_gpu.c'), pjoin(root_dir, 'prroi_pooling_gpu_impl.cu')],
verbose=True
)
except ImportError:
raise ImportError('Can not compile Precise RoI Pooling library.')
__all__ = ['prroi_pool2d']
class PrRoIPool2DFunction(ag.Function):
@staticmethod
def forward(ctx, features, rois, pooled_height, pooled_width, spatial_scale):
assert 'FloatTensor' in features.type() and 'FloatTensor' in rois.type(), \
'Precise RoI Pooling only takes float input, got {} for features and {} for rois.'.format(features.type(), rois.type())
pooled_height = int(pooled_height)