Spaces:
Running
on
Zero
Running
on
Zero
File size: 6,159 Bytes
a64b7d4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 |
# Modified from https://github.com/open-mmlab/mmcv/blob/master/mmcv/video/optflow.py # noqa: E501
import cv2
import numpy as np
import os
def flowread(flow_path, quantize=False, concat_axis=0, *args, **kwargs):
"""Read an optical flow map.
Args:
flow_path (ndarray or str): Flow path.
quantize (bool): whether to read quantized pair, if set to True,
remaining args will be passed to :func:`dequantize_flow`.
concat_axis (int): The axis that dx and dy are concatenated,
can be either 0 or 1. Ignored if quantize is False.
Returns:
ndarray: Optical flow represented as a (h, w, 2) numpy array
"""
if quantize:
assert concat_axis in [0, 1]
cat_flow = cv2.imread(flow_path, cv2.IMREAD_UNCHANGED)
if cat_flow.ndim != 2:
raise IOError(f'{flow_path} is not a valid quantized flow file, its dimension is {cat_flow.ndim}.')
assert cat_flow.shape[concat_axis] % 2 == 0
dx, dy = np.split(cat_flow, 2, axis=concat_axis)
flow = dequantize_flow(dx, dy, *args, **kwargs)
else:
with open(flow_path, 'rb') as f:
try:
header = f.read(4).decode('utf-8')
except Exception:
raise IOError(f'Invalid flow file: {flow_path}')
else:
if header != 'PIEH':
raise IOError(f'Invalid flow file: {flow_path}, header does not contain PIEH')
w = np.fromfile(f, np.int32, 1).squeeze()
h = np.fromfile(f, np.int32, 1).squeeze()
flow = np.fromfile(f, np.float32, w * h * 2).reshape((h, w, 2))
return flow.astype(np.float32)
def flowwrite(flow, filename, quantize=False, concat_axis=0, *args, **kwargs):
"""Write optical flow to file.
If the flow is not quantized, it will be saved as a .flo file losslessly,
otherwise a jpeg image which is lossy but of much smaller size. (dx and dy
will be concatenated horizontally into a single image if quantize is True.)
Args:
flow (ndarray): (h, w, 2) array of optical flow.
filename (str): Output filepath.
quantize (bool): Whether to quantize the flow and save it to 2 jpeg
images. If set to True, remaining args will be passed to
:func:`quantize_flow`.
concat_axis (int): The axis that dx and dy are concatenated,
can be either 0 or 1. Ignored if quantize is False.
"""
if not quantize:
with open(filename, 'wb') as f:
f.write('PIEH'.encode('utf-8'))
np.array([flow.shape[1], flow.shape[0]], dtype=np.int32).tofile(f)
flow = flow.astype(np.float32)
flow.tofile(f)
f.flush()
else:
assert concat_axis in [0, 1]
dx, dy = quantize_flow(flow, *args, **kwargs)
dxdy = np.concatenate((dx, dy), axis=concat_axis)
os.makedirs(os.path.dirname(filename), exist_ok=True)
cv2.imwrite(filename, dxdy)
def quantize_flow(flow, max_val=0.02, norm=True):
"""Quantize flow to [0, 255].
After this step, the size of flow will be much smaller, and can be
dumped as jpeg images.
Args:
flow (ndarray): (h, w, 2) array of optical flow.
max_val (float): Maximum value of flow, values beyond
[-max_val, max_val] will be truncated.
norm (bool): Whether to divide flow values by image width/height.
Returns:
tuple[ndarray]: Quantized dx and dy.
"""
h, w, _ = flow.shape
dx = flow[..., 0]
dy = flow[..., 1]
if norm:
dx = dx / w # avoid inplace operations
dy = dy / h
# use 255 levels instead of 256 to make sure 0 is 0 after dequantization.
flow_comps = [quantize(d, -max_val, max_val, 255, np.uint8) for d in [dx, dy]]
return tuple(flow_comps)
def dequantize_flow(dx, dy, max_val=0.02, denorm=True):
"""Recover from quantized flow.
Args:
dx (ndarray): Quantized dx.
dy (ndarray): Quantized dy.
max_val (float): Maximum value used when quantizing.
denorm (bool): Whether to multiply flow values with width/height.
Returns:
ndarray: Dequantized flow.
"""
assert dx.shape == dy.shape
assert dx.ndim == 2 or (dx.ndim == 3 and dx.shape[-1] == 1)
dx, dy = [dequantize(d, -max_val, max_val, 255) for d in [dx, dy]]
if denorm:
dx *= dx.shape[1]
dy *= dx.shape[0]
flow = np.dstack((dx, dy))
return flow
def quantize(arr, min_val, max_val, levels, dtype=np.int64):
"""Quantize an array of (-inf, inf) to [0, levels-1].
Args:
arr (ndarray): Input array.
min_val (scalar): Minimum value to be clipped.
max_val (scalar): Maximum value to be clipped.
levels (int): Quantization levels.
dtype (np.type): The type of the quantized array.
Returns:
tuple: Quantized array.
"""
if not (isinstance(levels, int) and levels > 1):
raise ValueError(f'levels must be a positive integer, but got {levels}')
if min_val >= max_val:
raise ValueError(f'min_val ({min_val}) must be smaller than max_val ({max_val})')
arr = np.clip(arr, min_val, max_val) - min_val
quantized_arr = np.minimum(np.floor(levels * arr / (max_val - min_val)).astype(dtype), levels - 1)
return quantized_arr
def dequantize(arr, min_val, max_val, levels, dtype=np.float64):
"""Dequantize an array.
Args:
arr (ndarray): Input array.
min_val (scalar): Minimum value to be clipped.
max_val (scalar): Maximum value to be clipped.
levels (int): Quantization levels.
dtype (np.type): The type of the dequantized array.
Returns:
tuple: Dequantized array.
"""
if not (isinstance(levels, int) and levels > 1):
raise ValueError(f'levels must be a positive integer, but got {levels}')
if min_val >= max_val:
raise ValueError(f'min_val ({min_val}) must be smaller than max_val ({max_val})')
dequantized_arr = (arr + 0.5).astype(dtype) * (max_val - min_val) / levels + min_val
return dequantized_arr
|