def test_forward_cpu_gpu_equal(self):
# cpu
x_cpu = chainer.Variable(self.x)
rois_cpu = chainer.Variable(self.rois)
roi_indices_cpu = chainer.Variable(self.roi_indices)
y_cpu = functions.roi_average_align_2d(
x_cpu,
rois_cpu,
roi_indices_cpu,
outsize=self.outsize,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio,
)
# gpu
x_gpu = chainer.Variable(cuda.to_gpu(self.x))
rois_gpu = chainer.Variable(cuda.to_gpu(self.rois))
roi_indices_gpu = chainer.Variable(cuda.to_gpu(self.roi_indices))
y_gpu = functions.roi_average_align_2d(
x_gpu,
rois_gpu,
roi_indices_gpu,
outsize=self.outsize,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio,
)
testing.assert_allclose(y_cpu.data, cuda.to_cpu(y_gpu.data))
def forward(self, hs, rois, roi_indices):
pooled_hs = []
for l, h in enumerate(hs):
if len(rois[l]) == 0:
continue
pooled_hs.append(
F.roi_average_align_2d(h, rois[l], roi_indices[l],
self._roi_size, self._scales[l],
self._roi_sample_ratio))
if len(pooled_hs) == 0:
out_size = self.segm_size
segs = chainer.Variable(
self.xp.empty((0, self._n_class, out_size, out_size),
dtype=np.float32))
return segs
h = F.concat(pooled_hs, axis=0)
h = self.conv1(h)
h = self.conv2(h)
h = self.conv3(h)
h = self.conv4(h)
h = F.relu(self.conv5(h))
return self.seg(h)
def f(x, rois, roi_indices):
return functions.roi_average_align_2d(
x,
rois,
roi_indices,
outsize=self.outsize,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio)
def f(x, rois):
return functions.roi_average_align_2d(
x,
rois,
outh=self.outh,
outw=self.outw,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio)
def test_forward_cpu_gpu_equal(self):
# cpu
x_cpu = chainer.Variable(self.x)
rois_cpu = chainer.Variable(self.rois)
y_cpu = functions.roi_average_align_2d(
x_cpu, rois_cpu, outh=self.outh, outw=self.outw,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio,
)
# gpu
x_gpu = chainer.Variable(cuda.to_gpu(self.x))
rois_gpu = chainer.Variable(cuda.to_gpu(self.rois))
y_gpu = functions.roi_average_align_2d(
x_gpu, rois_gpu, outh=self.outh, outw=self.outw,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio,
)
testing.assert_allclose(y_cpu.data, cuda.to_cpu(y_gpu.data))
def check_forward(self, x_data, roi_data):
x = chainer.Variable(x_data)
rois = chainer.Variable(roi_data)
y = functions.roi_average_align_2d(
x, rois, outh=self.outh, outw=self.outw,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio,
)
self.assertEqual(y.data.dtype, numpy.float32)
y_data = cuda.to_cpu(y.data)
self.assertEqual(self.gy.shape, y_data.shape)
def check_forward(self, x_data, roi_data, roi_index_data):
x = chainer.Variable(x_data)
rois = chainer.Variable(roi_data)
roi_indices = chainer.Variable(roi_index_data)
y = functions.roi_average_align_2d(
x, rois, roi_indices, outsize=self.outsize,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio,
)
self.assertEqual(y.data.dtype, numpy.float32)
y_data = cuda.to_cpu(y.data)
self.assertEqual(self.gy.shape, y_data.shape)
def forward(self, hs, rois, roi_indices):
"""Calculates RoIs.
Args:
hs (iterable of array): An iterable of feature maps.
rois (list of arrays): A list of arrays of shape: math: `(R_l, 4)`,
where: math: `R_l` is the number of RoIs in the: math: `l`- th
feature map.
roi_indices (list of arrays): A list of arrays of
shape :math:`(R_l,)`.
Returns:
tuple of two arrays:
:obj:`locs` and :obj:`confs`.
* **locs**: An arrays whose shape is \
:math:`(R, n\_class, 4)`, where :math:`R` is the total number \
of RoIs in the batch.
* **confs**: A list of array whose shape is :math:`(R, n\_class)`.
"""
hs_ = []
for l, h in enumerate(hs):
if len(rois[l]) == 0:
continue
h = F.roi_average_align_2d(h, rois[l], roi_indices[l],
self._roi_size, self._scales[l],
self._roi_sample_ratio)
hs_.append(h)
hs = hs_
if len(hs) == 0:
locs = chainer.Variable(
self.xp.empty((0, self._n_class, 4), dtype=np.float32))
confs = chainer.Variable(
self.xp.empty((0, self._n_class), dtype=np.float32))
return locs, confs
h = F.concat(hs, axis=0)
h = F.reshape(h, (h.shape[0], -1))
h = F.relu(self.fc1(h))
h = F.relu(self.fc2(h))
locs = self.loc(h)
locs = F.reshape(locs, (locs.shape[0], -1, 4))
confs = self.conf(h)
return locs, confs
def f(x, rois):
return functions.roi_average_align_2d(
x, rois, outh=self.outh, outw=self.outw,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio)
def f(x, rois, roi_indices):
return functions.roi_average_align_2d(
x, rois, roi_indices, outsize=self.outsize,
spatial_scale=self.spatial_scale,
sampling_ratio=self.sampling_ratio)