Spaces:
Runtime error
Runtime error
File size: 4,455 Bytes
5fd3fc6 | 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 | import glob
import os
import cv2
import numpy as np
import torch
from PIL import Image
from scipy.ndimage.morphology import binary_dilation
from ..dataset.range_transform import im_normalization
def images_to_torch(frames, device):
frames = torch.from_numpy(frames.transpose(0, 3, 1, 2)).float().unsqueeze(0) / 255
b, t, c, h, w = frames.shape
for ti in range(t):
frames[0, ti] = im_normalization(frames[0, ti])
return frames.to(device)
def load_images(path, min_side=None):
fnames = sorted(glob.glob(os.path.join(path, '*.jpg')))
if len(fnames) == 0:
fnames = sorted(glob.glob(os.path.join(path, '*.png')))
frame_list = []
for i, fname in enumerate(fnames):
if min_side:
image = Image.open(fname).convert('RGB')
w, h = image.size
new_w = (w * min_side // min(w, h))
new_h = (h * min_side // min(w, h))
frame_list.append(np.array(image.resize((new_w, new_h), Image.BICUBIC), dtype=np.uint8))
else:
frame_list.append(np.array(Image.open(fname).convert('RGB'), dtype=np.uint8))
frames = np.stack(frame_list, axis=0)
return frames
def load_masks(path, min_side=None):
fnames = sorted(glob.glob(os.path.join(path, '*.png')))
frame_list = []
first_frame = np.array(Image.open(fnames[0]))
binary_mask = (first_frame.max() == 255)
for i, fname in enumerate(fnames):
if min_side:
image = Image.open(fname)
w, h = image.size
new_w = (w * min_side // min(w, h))
new_h = (h * min_side // min(w, h))
frame_list.append(np.array(image.resize((new_w, new_h), Image.NEAREST), dtype=np.uint8))
else:
frame_list.append(np.array(Image.open(fname), dtype=np.uint8))
frames = np.stack(frame_list, axis=0)
if binary_mask:
frames = (frames > 128).astype(np.uint8)
return frames
def load_video(path, min_side=None):
frame_list = []
cap = cv2.VideoCapture(path)
while (cap.isOpened()):
_, frame = cap.read()
if frame is None:
break
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
if min_side:
h, w = frame.shape[:2]
new_w = (w * min_side // min(w, h))
new_h = (h * min_side // min(w, h))
frame = cv2.resize(frame, (new_w, new_h), interpolation=cv2.INTER_CUBIC)
frame_list.append(frame)
frames = np.stack(frame_list, axis=0)
return frames
def _pascal_color_map(N=256, normalized=False):
"""
Python implementation of the color map function for the PASCAL VOC data set.
Official Matlab version can be found in the PASCAL VOC devkit
http://host.robots.ox.ac.uk/pascal/VOC/voc2012/index.html#devkit
"""
def bitget(byteval, idx):
return (byteval & (1 << idx)) != 0
dtype = 'float32' if normalized else 'uint8'
cmap = np.zeros((N, 3), dtype=dtype)
for i in range(N):
r = g = b = 0
c = i
for j in range(8):
r = r | (bitget(c, 0) << 7 - j)
g = g | (bitget(c, 1) << 7 - j)
b = b | (bitget(c, 2) << 7 - j)
c = c >> 3
cmap[i] = np.array([r, g, b])
cmap = cmap / 255 if normalized else cmap
return cmap
color_map = [
[0, 0, 0],
[255, 50, 50],
[50, 255, 50],
[50, 50, 255],
[255, 50, 255],
[50, 255, 255],
[255, 255, 50],
]
color_map_np = np.array(color_map)
def overlay_davis(image, mask, alpha=0.5):
""" Overlay segmentation on top of RGB image. from davis official"""
im_overlay = image.copy()
colored_mask = color_map_np[mask]
foreground = image * alpha + (1 - alpha) * colored_mask
binary_mask = (mask > 0)
# Compose image
im_overlay[binary_mask] = foreground[binary_mask]
countours = binary_dilation(binary_mask) ^ binary_mask
im_overlay[countours, :] = 0
return im_overlay.astype(image.dtype)
def overlay_davis_fade(image, mask, alpha=0.5):
im_overlay = image.copy()
colored_mask = color_map_np[mask]
foreground = image * alpha + (1 - alpha) * colored_mask
binary_mask = (mask > 0)
# Compose image
im_overlay[binary_mask] = foreground[binary_mask]
countours = binary_dilation(binary_mask) ^ binary_mask
im_overlay[countours, :] = 0
im_overlay[~binary_mask] = im_overlay[~binary_mask] * 0.6
return im_overlay.astype(image.dtype)
|