def bbox3_to_mask(bbox3, shape):
"""
:param bbox3: [B, N, 4] y_from,x_from, y_size, x_size, normalized 0~1
:shape [H,W]
:return: [B, N, H, W] mask
"""
assert bbox3.ndim == 3 and bbox3.dims[-1] == 4
x, y = coord2d(shape) # [H, W]
x = x.expand_dims(0).expand_dims(0) # 1,1,H,W
y = y.expand_dims(0).expand_dims(0) # 1,1,H,W
bbox3 = bbox3.expand_dims(-1).expand_dims(-1)
xfrom = bbox3[:, :, 1] # B,N,1,1
xto = xfrom + bbox3[:, :, 3]
yfrom = bbox3[:, :, 0] # B,N,1,1
yto = yfrom + bbox3[:, :, 2]
# broadcasting
ix = tf.logical_and(x >= xfrom, x <= xto) # B,N,H,W
iy = tf.logical_and(y >= yfrom, y <= yto) # B,N,H,W
imask = tf.logical_and(ix, iy)
return imask
def bbox2_to_mask(bbox2, shape):
"""
:param bbox2: [N, 4] y_from,x_from, y_size, x_size, normalized 0~1
:param shape: (H, W)
:return: [N, H, W] mask
"""
# same to this
# return tf.map_fn(lambda x: bbox1_to_mask(x, shape), bbox3, dtype=tf.bool)
assert bbox2.ndim == 2 and bbox2.dims[-1] == 4
x, y = coord2d(shape) # H, W
x = x.expand_dims(0) # 1HW
y = y.expand_dims(0) # 1HW
bbox2 = bbox2.expand_dims(-1).expand_dims(-1)
xfrom = bbox2[:, 1] # N,1,1
xto = xfrom + bbox2[:, 3]
yfrom = bbox2[:, 0] # N,1,1
yto = yfrom + bbox2[:, 2]
ix = tf.logical_and(x >= xfrom, x <= xto)
iy = tf.logical_and(y >= yfrom, y <= yto)
imask = tf.logical_and(ix, iy)
return imask
def bbox1_to_mask(bbox1, shape):
"""
bbox
:param bbox1: [4] float32 value of (0~1) y_from,x_from, y_size, x_size, normalized 0~1
:param shape: [H,W]
:return: mask of dtype tf.bool
"""
# ex: shape = (255, 255)
# bbox = (N, 4)
assert bbox1.ndim == 1 and bbox1.dims[-1] == 4
x, y = coord2d(shape)
# check bbox x, bbox[1]~bbox[1]+bbox[3]
# ix = bbox[1] <= x <= (bbox[1] + bbox[3])
# iy = bbox[0] <= y <= (bbox[0] + bbox[2])
ix = tf.logical_and(x >= bbox1[1], x <= (bbox1[1] + bbox1[3]))
iy = tf.logical_and(y >= bbox1[0], y <= (bbox1[0] + bbox1[2]))
imask = tf.logical_and(ix, iy)
return imask
def sampling_xy_3r(img, xys, outsize=None, oob=None):
"""
differentiable image sampling (with interpolation)
:param img: source image [HWC]
:param xys: source coord [2, H'*W'] if outsize given
:param outsize: [H',W'] or None, xys must has rank3
:return: [B,H',W',C]
"""
assert img.ndim == 3
oobv = oob
if oobv is None:
# oobv = tf.zeros(shape=(img.dims[-1]), dtype=tf.float32) # [0., 0., 0.]
oobv = 0.
# oobv = [0., 0., 0.]
oobv = tf.convert_to_tensor(oobv)
if outsize is None:
outsize = tf.shape(xys)[1:]
xys = xys.flat2d()
H, W, C = img.shapes
WH = tf.stack([W, H]).to_float().reshape((2, 1))
# XYf = (xys + 1.) * WH * 0.5 # scale to HW coord ( + 1 for start from 0)
XYf = (xys +
0.5) * WH # * 0.5 # scale to HW coord ( + 1 for start from 0)
XYS = tf.ceil(XYf) # left top weight
# prepare weights
w00 = XYS - XYf # [2, p]
w11 = 1. - w00 # [2, p]
# get near 4 pixels per pixel
XYS = XYS.to_int32() # [2, p] # todo check xy order
XYs = XYS - 1
Xs = tf.stack([XYs[0], XYS[0]])
Ys = tf.stack([XYs[1], XYS[1]])
# get mask of outof bound
# leave option for filling value
Xi = Xs.clip_by_value(0, W - 1)
Yi = Ys.clip_by_value(0, H - 1)
inb = tf.logical_and(Xi.equal(Xs), Yi.equal(Ys)) # [2, p]
inb = tf.reduce_any(inb, axis=0, keepdims=True) # all oob? [1, p]-
# inb = inb.expand_dims(2).to_float() # [1, p]
inb = inb.reshape((-1, 1)).to_float() # [p, 1] 1 for channel
# get 4 pixels [p, C]
p00 = getpixel(img, tf.stack([Yi[0], Xi[0]]).T)
p01 = getpixel(img, tf.stack([Yi[0], Xi[1]]).T)
p10 = getpixel(img, tf.stack([Yi[1], Xi[0]]).T)
p11 = getpixel(img, tf.stack([Yi[1], Xi[1]]).T)
# stacked nearest : [4, p, C]
near4 = tf.stack([p00, p01, p10, p11], axis=0)
# XYw : 4 near point weights [4, pixel]
w4 = tf.stack([
w00[1] * w00[0], # left top
w00[1] * w11[0], # right top
w11[1] * w00[0], # left bottom
w11[1] * w11[0]
]) # right bottom
# weighted sum of 4 nearest pixels broadcasting
w4 = w4.reshape((4, -1, 1))
# interpolated = tf.sum(w4 * near4.to_float(), axis=1) # [p, C]
interpolated = tf.sum(w4 * near4.to_float(), axis=0) # [p, C]
# assign oob value
# fill oob by broadcasting
oobv = oobv.reshape((1, -1)) # [p, C]
interpolated = interpolated * inb + oobv * (1. - inb)
output = interpolated.reshape((outsize[0], outsize[1], C))
# reshape [p, C] => [H', W', C]
return output