def non_maximum_suppression(bbox, thresh, score=None, limit=None):
"""非极大值抑制"""
bbox_y1 = bbox[:, 0]
bbox_x1 = bbox[:, 1]
bbox_y2 = bbox[:, 2]
bbox_x2 = bbox[:, 3]
area = (bbox_x2 - bbox_x1 + 1) * (bbox_y2 - bbox_y1 + 1)
n_bbox = bbox.shape[0]
if score is not None:
order = score.argsort()[::-1].astype(np.int32)
else:
order = cp.arange(n_bbox, dtype=np.int32)
keep = []
# 预测框之间进行两两比较，去除重叠面积iou大于thresh的框
while order.size > 0:
i = order[0]
keep.append(i)
xx1 = cp.maximum(bbox_x1[i], bbox_x1[order[1:]])
yy1 = cp.maximum(bbox_y1[i], bbox_y1[order[1:]])
xx2 = cp.minimum(bbox_x2[i], bbox_x2[order[1:]])
yy2 = cp.minimum(bbox_y2[i], bbox_y2[order[1:]])
width = cp.maximum(0., (xx2 - xx1 + 1))
height = cp.maximum(0., (yy2 - yy1 + 1))
inter = width * height
iou = inter / (area[i] + area[order[1:]] - inter)
index = cp.where(iou <= thresh)[0]
order = order[(index + 1).tolist()]
if limit is not None:
keep = keep[:limit]
return cp.asnumpy(keep)
def _get_nearplane_steps(diff, dOP, dPO, A1, A4, recover_psi, recover_probe):
# (22) Use least-squares to find the optimal step sizes simultaneously
if recover_psi and recover_probe:
b1 = cp.sum((dOP.conj() * diff).real, axis=(-2, -1))
b2 = cp.sum((dPO.conj() * diff).real, axis=(-2, -1))
A2 = cp.sum((dOP * dPO.conj()), axis=(-2, -1))
A3 = A2.conj()
determinant = A1 * A4 - A2 * A3
x1 = -cp.conj(A2 * b2 - A4 * b1) / determinant
x2 = cp.conj(A1 * b2 - A3 * b1) / determinant
elif recover_psi:
b1 = cp.sum((dOP.conj() * diff).real, axis=(-2, -1))
x1 = b1 / A1
elif recover_probe:
b2 = cp.sum((dPO.conj() * diff).real, axis=(-2, -1))
x2 = b2 / A4
if recover_psi:
step = 0.9 * cp.maximum(0, x1[..., None, None].real)
# (27b) Object update
weighted_step_psi = cp.mean(step, keepdims=True, axis=-5)
if recover_probe:
step = 0.9 * cp.maximum(0, x2[..., None, None].real)
weighted_step_probe = cp.mean(step, axis=-5, keepdims=True)
else:
weighted_step_probe = None
return weighted_step_psi, weighted_step_probe
def non_max_suppression(self, A, threshold=0.5):
"""Remove overlapping bounding boxes. Returns filterd boxes in screen coordinates and
as an array with shape (n_boxes, (x1, y1, x2, y2)).
Args:
A (np.array): predicted labels and boxes.
threshold (float): overlap threshold to treat as new box
Returns:
np.array: only max boxes.
"""
x_stride = 34
y_stride = 34
score = []
x1 = []
x2 = []
y1 = []
y2 = []
for i in range(0, A.shape[1]):
for j in range(0, A.shape[2]):
if A[0, i, j][0] > 0.5:
bx, by, w, h = A[0, i, j][1:]
score.append(A[0, i, j, 0])
x1.append((j + bx - w / 2) * x_stride + 29)
y1.append((i + by - h / 2) * y_stride)
x2.append((j + bx + w / 2) * x_stride + 29)
y2.append((i + by + h / 2) * y_stride)
score = np.array(score)
x1 = cp.array(x1)
x2 = cp.array(x2)
y1 = cp.array(y1)
y2 = cp.array(y2)
score_indexes = score.argsort().tolist()
boxes_keep_index = []
while len(score_indexes) > 0:
index = score_indexes.pop()
boxes_keep_index.append(index)
if not len(score_indexes):
break
#iou
xs1 = cp.maximum(x1[index], x1[score_indexes])
ys1 = cp.maximum(y1[index], y1[score_indexes])
xs2 = cp.minimum(x2[index], x2[score_indexes])
ys2 = cp.minimum(y2[index], y2[score_indexes])
intersections = cp.maximum(ys2 - ys1, 0) * cp.maximum(xs2 - xs1, 0)
unions = (x2[index]-x1[index])*(y2[index]-y1[index]) \
+ (x2[score_indexes]-x1[score_indexes])*(y2[score_indexes]-y1[score_indexes]) \
- intersections
ious = np.array(cp.asnumpy(intersections / unions))
filtered_indexes = set((ious > threshold).nonzero()[0])
score_indexes = [
v for (i, v) in enumerate(score_indexes)
if i not in filtered_indexes
]
nms_res = np.zeros((len(boxes_keep_index), 5))
for i, j in enumerate(boxes_keep_index):
nms_res[i, :] = np.array([score[j], x1[j], y1[j], x2[j], y2[j]])
return nms_res
def calcDifference(sample, aNeg, bNeg, aPos, bPos):
negPDF = beta.pdf(sample, aNeg, bNeg)
posPDF = beta.pdf(sample, aPos, bPos)
pdfDiff = negPDF - posPDF
pdfDiffNeg = xp.maximum(pdfDiff, xp.zeros_like(pdfDiff))
pdfDiffPos = xp.maximum(-1 * pdfDiff, xp.zeros_like(pdfDiff))
pdfMax = xp.maximum(negPDF, posPDF)
return negPDF, posPDF, pdfDiffPos, pdfDiffNeg, pdfMax
def stretch_pre(nimg):
"""
from 'Applicability Of White-Balancing Algorithms to Restoring Faded Colour Slides: An Empirical Evaluation'
"""
nimg = nimg.transpose(2, 0, 1)
nimg[0] = np.maximum(nimg[0] - nimg[0].min(), 0)
nimg[1] = np.maximum(nimg[1] - nimg[1].min(), 0)
nimg[2] = np.maximum(nimg[2] - nimg[2].min(), 0)
return nimg.transpose(1, 2, 0)
def prox(self, x):
if self.prox_method == 'tv':
x = 0.5 * (np.maximum(x, 0) + tv.tv3dApproxHaar(
x, self.tv_lambda / self.L, self.tv_lambdaw))
if self.prox_method == 'native':
x = np.maximum(x, 0) + self.soft_thresh(x, self.tau)
if self.prox_method == 'non-neg':
x = np.maximum(x, 0)
return x
def _calc_offset(offset_map, region, parent_y, parent_x):
y0, y1, x0, x1 = region
x_area, y_area = offset_map[:, y0:y1, x0:x1]
x_vec = np.power(np.subtract(np.arange(x0, x1), parent_x), 2)
y_vec = np.power(np.subtract(np.arange(y0, y1), parent_y), 2)
xv, yv = np.meshgrid(x_vec, y_vec)
dist = np.sqrt(xv + yv) # sqrt(y^2 + x^2)
xv = np.divide(xv, dist) # normalize x
yv = np.divide(yv, dist) # normalize y
offset_map[0, y0:y1, x0:x1] = np.maximum(x_area, xv)
offset_map[1, y0:y1, x0:x1] = np.maximum(y_area, yv)
def yangVectorDistance(negativeVector, positiveVector, p=1):
x = xp.array(negativeVector).reshape((-1, 1))
y = xp.array(positiveVector).reshape((-1, 1))
pExp = int(p)
assert x.shape == y.shape, "x ({}) and y ({}) must be of the same shape".format(
x.shape, y.shape)
assert pExp > 0, "p must be an integer greater than 0"
numerator = vectorPDistance(x, y, pExp)
max_X_Y = xp.maximum(xp.absolute(x), xp.absolute(y))
maxes = xp.maximum(max_X_Y, xp.absolute(x - y))
return numerator / xp.sum(maxes)
def fast_forward_one(
prev_x: cp.ndarray,
prev_l: cp.ndarray,
hidden: cp.ndarray,
x_embedder_W: cp.ndarray,
gru_xw: cp.ndarray,
gru_hw: cp.ndarray,
gru_xb: cp.ndarray,
gru_hb: cp.ndarray,
O1_W: cp.ndarray,
O1_b: cp.ndarray,
O2_W: cp.ndarray,
O2_b: cp.ndarray,
w_gru_x: cp.ndarray,
w_gru_h: cp.ndarray,
w_out_x1: cp.ndarray,
w_out_x2: cp.ndarray,
):
prev_xl = cp.concatenate((x_embedder_W[prev_x], prev_l),
axis=1) # (batch_size, ?)
# gru_x = prev_xl.dot(gru_xw) + gru_xb
gru_x = w_gru_x
prev_xl.dot(gru_xw, gru_x)
gru_x += gru_xb
# gru_h = hidden.dot(gru_hw) + gru_hb
gru_h = w_gru_h
hidden.dot(gru_hw, gru_h)
gru_h += gru_hb
size = gru_x.shape[1] // 3
W_r_x, W_z_x, W_x = gru_x[:, :size], gru_x[:,
size:size * 2], gru_x[:,
size * 2:]
U_r_h, U_z_h, U_x = gru_h[:, :size], gru_h[:,
size:size * 2], gru_h[:,
size * 2:]
new_hidden = gru_element_wise(hidden, W_r_x, W_z_x, W_x, U_r_h, U_z_h, U_x)
# out_x = new_hidden.dot(O1_W) + O1_b
out_x1 = w_out_x1
new_hidden.dot(O1_W, out_x1)
out_x1 += O1_b
cp.maximum(out_x1, 0.0, out_x1)
# out_x = out_x.dot(O2_W) + O2_b
out_x2 = w_out_x2
out_x1.dot(O2_W, out_x2)
out_x2 += O2_b
return out_x2, new_hidden
def _update_position(
scan,
position_options,
position_update_numerator,
position_update_denominator,
alpha=0.05,
max_shift=1,
):
step = position_update_numerator / (
(1 - alpha) * position_update_denominator +
alpha * position_update_denominator.max())
step_x = step[..., 0]
step_y = step[..., 1]
if position_options.use_adaptive_moment:
logger.info(
"position correction with ADAptive Momemtum acceleration enabled.")
step_x, position_options.vx, position_options.mx = adam(
step_x,
position_options.vx,
position_options.mx,
vdecay=position_options.vdecay,
mdecay=position_options.mdecay)
step_y, position_options.vy, position_options.my = adam(
step_y,
position_options.vy,
position_options.my,
vdecay=position_options.vdecay,
mdecay=position_options.mdecay)
# Step limit for stability
_max_shift = cp.minimum(
max_shift,
_mad(
cp.concatenate((step_x, step_y), axis=-1),
axis=-1,
keepdims=True,
),
)
step_x = cp.maximum(-_max_shift, cp.minimum(step_x, _max_shift))
step_y = cp.maximum(-_max_shift, cp.minimum(step_y, _max_shift))
# Ensure net movement is zero
step_x -= cp.mean(step_x, axis=-1, keepdims=True)
step_y -= cp.mean(step_y, axis=-1, keepdims=True)
logger.info('position update norm is %+.3e', tike.linalg.norm(step_x))
scan[..., 0] -= step_x
scan[..., 1] -= step_y
return scan, position_options
def test_max(self):
@jit.rawkernel()
def f(x, y, z, r):
tid = jit.blockDim.x * jit.blockIdx.x + jit.threadIdx.x
r[tid] = max(x[tid], y[tid], z[tid])
x = testing.shaped_random((1024,), dtype=numpy.int32, seed=0)
y = testing.shaped_random((1024,), dtype=numpy.int32, seed=1)
z = testing.shaped_random((1024,), dtype=numpy.int32, seed=2)
r = testing.shaped_random((1024,), dtype=numpy.int32, seed=3)
f((8,), (128,), (x, y, z, r))
expected = cupy.maximum(x, cupy.maximum(y, z))
assert bool((r == expected).all())
def normal_density_cupy(x, mean, stddev, from_axis=None, eps=1e-8, gpu=0):
import cupy as cp
with cp.cuda.Device(gpu):
variance = cp.maximum(stddev ** 2, eps)
stddev = cp.maximum(stddev, eps)
density = cp.exp(-cp.square(x - mean) / (2 * variance)) / (stddev * math.sqrt(2 * math.pi))
if (from_axis is not None) and (from_axis >= 0):
shape = tuple(density.shape[:from_axis]) + (cp.prod(density.shape[from_axis:]),)
density = cp.reshape(density, shape)
density = cp.prod(density, axis=from_axis)
return density
def normal_log_density_cupy(x, mean, stddev, from_axis=None, eps=1e-8, gpu=0):
import cupy as cp
with cp.cuda.Device(gpu):
variance = cp.maximum(stddev ** 2, eps)
log_stddev = cp.log(cp.maximum(stddev, eps))
log_density = -0.5 * (math.log(2 * math.pi) + 2 * log_stddev + ((x - mean)**2 / variance))
if (from_axis is not None) and (from_axis >= 0):
shape = tuple(log_density.shape[:from_axis]) + (cp.prod(log_density.shape[from_axis:]),)
log_density = cp.reshape(log_density, shape)
log_density = cp.sum(log_density, axis=from_axis)
return log_density
def _nms_boxes(self, boxes, box_confidences):
"""Apply the Non-Maximum Suppression (NMS) algorithm on the bounding boxes with their
confidence scores and return an array with the indexes of the bounding boxes we want to
keep (and display later).
Keyword arguments:
boxes -- a NumPy array containing N bounding-box coordinates that survived filtering,
with shape (N,4); 4 for x,y,height,width coordinates of the boxes
box_confidences -- a Numpy array containing the corresponding confidences with shape N
"""
x_coord = boxes[:, 0]
y_coord = boxes[:, 1]
width = boxes[:, 2]
height = boxes[:, 3]
areas = width * height
ordered = box_confidences.argsort()[::-1]
keep = list()
while ordered.size > 0:
# Index of the current element:
i = ordered[0]
ii = cp.asnumpy(i)
keep.append(ii)
xx1 = cp.maximum(x_coord[i], x_coord[ordered[1:]])
yy1 = cp.maximum(y_coord[i], y_coord[ordered[1:]])
xx2 = cp.minimum(x_coord[i] + width[i],
x_coord[ordered[1:]] + width[ordered[1:]])
yy2 = cp.minimum(y_coord[i] + height[i],
y_coord[ordered[1:]] + height[ordered[1:]])
width1 = cp.maximum(0.0, xx2 - xx1 + 1)
height1 = cp.maximum(0.0, yy2 - yy1 + 1)
intersection = width1 * height1
union = (areas[i] + areas[ordered[1:]] - intersection)
# Compute the Intersection over Union (IoU) score:
iou = intersection / union
# The goal of the NMS algorithm is to reduce the number of adjacent bounding-box
# candidates to a minimum. In this step, we keep only those elements whose overlap
# with the current bounding box is lower than the threshold:
indexes = cp.where(iou <= self.nms_threshold)[0]
ordered = ordered[indexes + 1]
keep = np.array(keep)
print(keep)
keep = cp.asarray(keep)
return keep
def compute_gain(sound, fs, min_db=-80.0, mode='A_weighting'):
if fs == 16000:
n_fft = 2048
elif fs == 44100:
n_fft = 4096
else:
raise Exception('Invalid fs {}'.format(fs))
stride = n_fft // 2
gain = None
for i in range(0, len(sound[0]) - n_fft + 1, stride):
if mode == 'RMSE':
g = cupy.mean(sound[i:i + n_fft]**2, axis=1)
elif mode == 'A_weighting':
spec = cupy.fft.rfft(
cupy.hanning(n_fft + 1)[:-1] * sound[:, i:i + n_fft])
power_spec = cupy.abs(spec)**2
a_weighted_spec = power_spec * cupy.power(10,
a_weight(fs, n_fft) / 10)
g = cupy.sum(a_weighted_spec, axis=1)
else:
raise Exception('Invalid mode {}'.format(mode))
if i == 0:
gain = g.reshape([-1, 1])
else:
gain = cupy.concatenate((gain, g.reshape([-1, 1])), axis=1)
gain = cupy.maximum(gain, cupy.power(10, min_db / 10))
gain_db = 10 * cupy.log10(gain)
return gain_db
def adam(params, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, amsgrad, beta1, beta2, lr, weight_decay, eps):
_check_tensors(*params)
engine = _get_engine(*params)
for i, param in enumerate(params):
grad = grads[i]
exp_avg = exp_avgs[i]
exp_avg_sq = exp_avg_sqs[i]
step = state_steps[i]
bias_correction1 = 1 - beta1 ** step
bias_correction2 = 1 - beta2 ** step
if weight_decay != 0:
grad._data = grad.data + param.data * weight_decay
# Decay the first and second moment running average coefficient
exp_avg._data = exp_avg.data * beta1 + (1 - beta1) * grad.data
# check if this is true
exp_avg_sq._data = exp_avg_sq.data * beta2
exp_avg_sq._data = exp_avg_sq.data + (1 - beta2) * (grad.data * grad.data)
# exp_avg_sq._data = exp_avg_sq.data * beta2 + (1 - beta2) * (grad.data * grad.data)
if amsgrad:
max_exp_avg_sq = max_exp_avg_sqs[i]
# Maintains the maximum of all 2nd moment running avg. till now
max_exp_avg_sq._data = engine.maximum(max_exp_avg_sq.data, exp_avg_sq.data)
# Use the max. for normalizing running avg. of gradient
denom = (engine.sqrt(max_exp_avg_sq.data) / math.sqrt(bias_correction2)) + eps
else:
denom = (engine.sqrt(exp_avg_sq.data) / math.sqrt(bias_correction2)) + eps
step_size = lr / bias_correction1
param._data = param.data - step_size * (exp_avg.data / denom)
def mean_peak_distance(peak_image, centroids, return_numpy=True):
"""
Calculate the mean peak distance in degrees between two corresponding peaks
for each line profile in an SLI image series.
Args:
peak_image: Boolean NumPy array specifying the peak positions in the full
SLI stack
centroids: Use centroid calculation to better determine the peak position
regardless of the number of
measurements / illumination angles used.
return_numpy: Necessary if using `use_gpu`. Specifies if a CuPy or Numpy
array will be returned.
Returns:
NumPy array of floating point values containing the mean peak distance of
the line profiles in degrees.
"""
peak_distance_gpu = peak_distance(peak_image, centroids,
return_numpy=False)
peak_distance_gpu[peak_distance_gpu > 180] = 0
peak_distance_gpu = cupy.sum(peak_distance_gpu, axis=-1) / \
cupy.maximum(1, cupy.count_nonzero(peak_distance_gpu,
axis=-1))
if return_numpy:
peak_width_cpu = cupy.asnumpy(peak_distance_gpu)
del peak_distance_gpu
return peak_width_cpu
else:
return peak_distance_gpu
def maximum(a, cuda=False):
if cuda:
res = cp.maximum(a, 0)
cp.cuda.Stream.null.synchronize()
return res
else:
return np.maximum(a, 0)
def FSITM(HDR, LDR, alpha=None):
NumPixels = LDR.size
if alpha is None:
r = cp.floor(NumPixels / (2.**18))
if r > 1.:
alpha = 1. - (1. / r)
else:
alpha = 0.
minNonzero = cp.min(HDR[HDR > 0])
LogH = cp.log(cp.maximum(HDR, minNonzero))
# float is needed for further calculation
LogH = cp.around((LogH - LogH.min()) * 255. /
(LogH.max() - LogH.min())).astype(cp.float)
if alpha > 0.:
PhaseHDR_CH = phasecong100(HDR, 2, 2, 8, 8)
PhaseLDR_CH8 = phasecong100(LDR, 2, 2, 8, 8)
else: # so, if image size is smaller than 512x512?
PhaseHDR_CH = 0
PhaseLDR_CH8 = 0
PhaseLogH = phasecong100(LogH, 2, 2, 2, 2)
PhaseH = alpha * PhaseHDR_CH + (1 - alpha) * PhaseLogH
PhaseLDR_CH2 = phasecong100(LDR, 2, 2, 2, 2)
PhaseL = alpha * PhaseLDR_CH8 + (1 - alpha) * PhaseLDR_CH2
Q = cp.sum(
cp.logical_or(cp.logical_and(PhaseL <= 0, PhaseH <= 0),
cp.logical_and(PhaseL > 0, PhaseH > 0))) / NumPixels
return Q
def ruzicka_mat(matrix_a, vector_new):
matrix_a *= cp.arange(1023, -1, -1, dtype=cp.uint16)
min_up = cp.minimum(cp.array(matrix_a), vector_new)
max_down = cp.maximum(cp.array(matrix_a), vector_new)
numerator = cp.sum(min_up, axis=1)
denominator = cp.sum(max_down, axis=1)
return cp.asnumpy(cp.divide(numerator, denominator))
def relu(self, Z):
"""ReLU function"""
A = cp.maximum(0, Z)
assert (A.shape == Z.shape)
return A
def _min_or_max(self, axis, out, min_or_max, sum_duplicates, non_zero):
if out is not None:
raise ValueError(("Sparse matrices do not support "
"an 'out' parameter."))
util.validateaxis(axis)
if axis is None:
if 0 in self.shape:
raise ValueError("zero-size array to reduction operation")
zero = cupy.zeros((), dtype=self.dtype)
if self.nnz == 0:
return zero
if sum_duplicates:
self.sum_duplicates()
m = min_or_max(self.data)
if non_zero:
return m
if self.nnz != internal.prod(self.shape):
if min_or_max is cupy.min:
m = cupy.minimum(zero, m)
elif min_or_max is cupy.max:
m = cupy.maximum(zero, m)
else:
assert False
return m
if axis == 0 or axis == 1:
return self._min_or_max_axis(axis, min_or_max, sum_duplicates,
non_zero)
else:
raise ValueError("axis out of range")
def smearing(self, f, f_el, volt_mat, new_ind):
'''
Produces B matrix by comparing voltages
takes:
f - array shape (n_nodes)
f_el - array shape (n_electrodes)
volt_mat - array shape (n_measurements, 2)
new_ind - array shape (n_measurements)
returns:
b-matrix - array shape (n_measurements, n_nodes)
'''
i = cp.arange(len(volt_mat))
f_volt0 = f_el[new_ind, volt_mat[:, 0].astype(int)]
f_volt1 = f_el[new_ind, volt_mat[:, 1].astype(int)]
min_fel = cp.minimum(f_volt0, f_volt1)
max_fel = cp.maximum(f_volt0, f_volt1)
b_matrix = cp.empty((len(volt_mat), self.n_pts+self.ne))
b_matrix[:] = (min_fel[:, None] < f[new_ind]) & (f[new_ind] <= max_fel[:, None])
return b_matrix
def _min_or_max(self, axis, out, min_or_max, explicit):
if out is not None:
raise ValueError(("Sparse matrices do not support "
"an 'out' parameter."))
sputils.validateaxis(axis)
if axis is None:
if 0 in self.shape:
raise ValueError("zero-size array to reduction operation")
zero = cupy.zeros((), dtype=self.dtype)
if self.nnz == 0:
return zero
self.sum_duplicates()
m = min_or_max(self.data)
if explicit:
return m
if self.nnz != internal.prod(self.shape):
if min_or_max is cupy.min:
m = cupy.minimum(zero, m)
elif min_or_max is cupy.max:
m = cupy.maximum(zero, m)
else:
assert False
return m
if axis < 0:
axis += 2
return self._min_or_max_axis(axis, min_or_max, explicit)
def relu(z, a=None, derivative=False):
if derivative:
return z > 0
else:
# z[z<0]=0
# return z
# return z*(z>0)
return cp.maximum(0, z)
def _relu(self, z):
"""ReLu activation function
Args:
z (np.array): input
Returns:
np.array.
"""
return np.maximum(0, z)
def lrelu(x, alpha=0.01, derivative=False):
res = x
if derivative:
dx = np.ones_like(res)
dx[res < 0] = alpha
return dx
else:
return np.maximum(x, x * alpha, x)
def ruzicka_vec(vector_old, vector_new):
vector_old_cp = cp.array(vector_old) * cp.arange(
1023, -1, -1, dtype=cp.uint16)
min_up = cp.minimum(vector_old_cp, vector_new)
max_down = cp.maximum(vector_old_cp, vector_new)
numerator = cp.sum(min_up, axis=1)
denominator = cp.sum(max_down, axis=1)
return cp.asnumpy(cp.divide(numerator, denominator))
def darkflat_correction(self, data):
"""Dark-flat field correction"""
for k in range(
data.shape[0]): # work with 2D arrays to save GPU memory
data[k] = (data[k] - self.dark) / cp.maximum(
self.flat - self.dark, 1e-6)
return data
def predict(u):
x = u
for weight, bias in zip(weights, biases):
x = np.matmul(x, weight)
x = np.add(x, bias)
x = np.maximum(0, x)
return x
def _fftconv(a, b, axes=(0, 1)):
"""Patched version of :func:`sporco.linalg.fftconv`."""
if cp.isrealobj(a) and cp.isrealobj(b):
fft = cp.fft.rfftn
ifft = cp.fft.irfftn
else:
fft = cp.fft.fftn
ifft = cp.fft.ifftn
dims = cp.maximum(cp.asarray([a.shape[i] for i in axes]),
cp.asarray([b.shape[i] for i in axes]))
dims = [int(d) for d in dims]
af = fft(a, dims, axes)
bf = fft(b, dims, axes)
return ifft(af * bf, dims, axes)
