File size: 7,624 Bytes
3cf4fff |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 |
import os
import glob
import cv2
from tqdm import tqdm
from torchvision.transforms import (
Compose,
Resize,
CenterCrop,
ToTensor,
Normalize,
InterpolationMode,
)
from PIL import Image
import torch
import numpy as np
import pandas as pd
def convert_avi_to_mp4(input_path, output_path):
# Check if output_path already exists
# Convert .avi to .mp4 using OpenCV
cap = cv2.VideoCapture(input_path)
if not cap.isOpened():
raise IOError(f"Cannot open {input_path}")
fourcc = cv2.VideoWriter_fourcc(*"mp4v") # Use mp4 codec
fps = cap.get(cv2.CAP_PROP_FPS)
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
out = cv2.VideoWriter(output_path, fourcc, fps, (width, height))
while True:
ret, frame = cap.read()
if not ret:
break
out.write(frame)
cap.release()
out.release()
def create_images(video_path, output_dir):
if os.path.isdir(output_dir):
pass
else:
os.makedirs(output_dir, exist_ok=True)
cap = cv2.VideoCapture(video_path)
frame_idx = 0
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
frame_path = os.path.join(output_dir, f"{frame_idx:05d}.jpg")
cv2.imwrite(frame_path, frame)
frame_idx += 1
def create_videos(video_path, output_dir):
# Determine input and output paths for .avi and .webm
base_path = video_path
mp4_path = video_path.replace("/Data/", "/Video_data/")
if mp4_path.endswith(".avi") or mp4_path.endswith(".webm"):
mp4_path = mp4_path.rsplit(".", 1)[0] + ".mp4"
else:
mp4_path = mp4_path.replace(".mp4", ".mp4") # fallback, should already be .mp4
os.makedirs(output_dir, exist_ok=True)
if not os.path.exists(mp4_path):
# Try .avi first
avi_path = video_path.replace(".mp4", ".avi")
if os.path.exists(avi_path):
convert_avi_to_mp4(avi_path, mp4_path)
else:
# Try .webm
webm_path = video_path.replace(".mp4", ".webm")
if os.path.exists(webm_path):
convert_avi_to_mp4(webm_path, mp4_path)
else:
raise FileNotFoundError(
f"Neither .avi nor .webm found for {video_path}"
)
# Code to convert one video to few images.
def video2image(video_path, frame_rate=1.0, size=224):
def preprocess(size, n_px):
return Compose(
[
Resize(size, interpolation=InterpolationMode.BICUBIC),
CenterCrop(size),
lambda image: image.convert("RGB"),
ToTensor(),
Normalize(
(0.48145466, 0.4578275, 0.40821073),
(0.26862954, 0.26130258, 0.27577711),
),
]
)(n_px)
cap = cv2.VideoCapture(video_path)
cap = cv2.VideoCapture(video_path, cv2.CAP_FFMPEG)
frameCount = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
fps = int(cap.get(cv2.CAP_PROP_FPS))
if fps < 1:
images = np.zeros([3, size, size], dtype=np.float32)
print("ERROR: problem reading video file: ", video_path)
else:
total_duration = (frameCount + fps - 1) // fps
start_sec, end_sec = 0, total_duration
interval = fps / frame_rate
frames_idx = np.floor(np.arange(start_sec * fps, end_sec * fps, interval))
ret = True
images = np.zeros([len(frames_idx), 3, size, size], dtype=np.float32)
for i, idx in enumerate(frames_idx):
cap.set(cv2.CAP_PROP_POS_FRAMES, idx)
ret, frame = cap.read()
if not ret:
break
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
last_frame = i
images[i, :, :, :] = preprocess(size, Image.fromarray(frame).convert("RGB"))
images = images[: last_frame + 1]
cap.release()
video_frames = torch.tensor(images)
return video_frames
## Dataset specific
def create_images_breakfast(gobal_path, start_end=15):
sub_dirs = sorted(glob.glob(gobal_path + "*/"))
def get_folder_name(local_path):
return local_path.split("/")[-2].replace("P", "")
part_dirs = []
for i in sub_dirs:
if int(get_folder_name(i)) <= start_end:
part_dirs.append(i)
for i in tqdm(part_dirs, desc="Processing avi to images"):
video_dirs = sorted(glob.glob(i + "*/"))
for j in video_dirs:
avi_files = sorted(glob.glob(j + "*.avi"))
for k in avi_files:
output_dir = k.replace("/Data/", "/Image_data/").replace(".avi", "")
create_images(k, output_dir)
def create_videos_breakfast(gobal_path, start_end=15):
sub_dirs = sorted(glob.glob(gobal_path + "*/"))
def get_folder_name(local_path):
return local_path.split("/")[-2].replace("P", "")
part_dirs = []
for i in sub_dirs:
if int(get_folder_name(i)) <= start_end:
part_dirs.append(i)
for i in tqdm(part_dirs, desc="Processing avi to mp4"):
video_dirs = sorted(glob.glob(i + "*/"))
for j in video_dirs:
avi_files = sorted(glob.glob(j + "*.avi"))
for k in avi_files:
output_dir = j.replace("/Data/", "/Video_data/")
create_videos(k, output_dir)
def create_images_ucf(global_path, files):
path_list = pd.read_csv(files, sep=" ", header=None)
for i in tqdm(path_list.values):
video_path = os.path.join(global_path, i[0])
output_dir = video_path.replace("/Data/", "/Image_data/").replace(".avi", "")
create_images(video_path, output_dir)
def create_videos_ucf(global_path, files):
path_list = pd.read_csv(files, sep=" ", header=None)
for i in tqdm(path_list.values):
video_path = os.path.join(global_path, i[0])
output_dir = os.path.dirname(
video_path.replace("/Data/", "/Video_data/").replace(".avi", "")
)
create_videos(video_path, output_dir)
def create_images_hmdb(global_path, path_list):
for i in tqdm(path_list):
local_name = i.split("/")[1]
video_path = global_path + i
output_dir = (
video_path.replace("/Data/", "/Image_data/")
.replace(".avi", "")
.replace("//", "/")
)
video_path = global_path + local_name + i
create_images(video_path, output_dir)
def create_videos_hmdb(global_path, path_list):
for i in tqdm(path_list):
local_name = i.split("/")[1]
video_path = global_path + local_name + i
output_dir = (
video_path.replace("/Data/", "/Video_data/")
.replace(".avi", "")
.replace("//", "/")
)
# remove last folder in outputdir
output_dir = os.path.dirname(output_dir)
video_path = global_path + local_name + i
create_videos(video_path, output_dir)
def create_images_sth2(global_path, files):
for i in tqdm(files):
video_path = global_path + str(i) + ".webm"
output_dir = video_path.replace("/Data/", "/Image_data/").replace(".webm", "")
create_images(video_path, output_dir)
def create_videos_sth2(global_path, files):
for i in tqdm(files):
video_path = global_path + str(i) + ".webm"
output_dir = os.path.dirname(
video_path.replace("/Data/", "/Video_data/").replace(".webm", "")
)
create_videos(video_path, output_dir)
|