How to use decord - 10 common examples

def read_data(opt, video_name, transform):

    decord_vr = decord.VideoReader(video_name, width=opt.new_width, height=opt.new_height)
    duration = len(decord_vr)

    opt.skip_length = opt.new_length * opt.new_step
    segment_indices, skip_offsets = sample_indices(opt, duration)

    if opt.video_loader:
        if opt.slowfast:
            clip_input = video_TSN_decord_slowfast_loader(opt, video_name, decord_vr, duration, segment_indices, skip_offsets)
        else:
            clip_input = video_TSN_decord_batch_loader(opt, video_name, decord_vr, duration, segment_indices, skip_offsets)

    clip_input = transform(clip_input)

    if opt.slowfast:
        sparse_sampels = len(clip_input) // (opt.num_segments * opt.num_crop)
        clip_input = np.stack(clip_input, axis=0)

def read_data(opt, video_name, transform):

    decord_vr = decord.VideoReader(video_name, width=opt.new_width, height=opt.new_height)
    duration = len(decord_vr)

    opt.skip_length = opt.new_length * opt.new_step
    segment_indices, skip_offsets = sample_indices(opt, duration)

    if opt.video_loader:
        if opt.slowfast:
            clip_input = video_TSN_decord_slowfast_loader(opt, video_name, decord_vr, duration, segment_indices, skip_offsets)
        else:
            clip_input = video_TSN_decord_batch_loader(opt, video_name, decord_vr, duration, segment_indices, skip_offsets)

    clip_input = transform(clip_input)

    if opt.slowfast:
        sparse_sampels = len(clip_input) // (opt.num_segments * opt.num_crop)
        clip_input = np.stack(clip_input, axis=0)

parser.add_argument('--gpu', type=int, default=-1, help='context to run, use --gpu=-1 to use cpu only')
parser.add_argument('--file', type=str, default='/tmp/testsrc_h264_100s_default.mp4', help='Test video')
parser.add_argument('--seed', type=int, default=666, help='numpy random seed for random access indices')
parser.add_argument('--random-frames', type=int, default=300, help='number of random frames to run')
parser.add_argument('--width', type=int, default=320, help='resize frame width')
parser.add_argument('--height', type=int, default=240, help='resize frame height')

args = parser.parse_args()

test_video = args.file
if args.gpu > -1:
    ctx = de.gpu(args.gpu)
else:
    ctx = de.cpu()

vr = de.VideoReader(test_video, ctx, width=args.width, height=args.height)
cnt = 0
tic = time.time()
while True:
    try:
        frame = vr.next()
    except StopIteration:
        break
    cnt += 1
print(cnt, ' frames, elapsed time for sequential read: ', time.time() - tic)

np.random.seed(args.seed)  # fix seed for all random tests
acc_indices = np.arange(len(vr))
np.random.shuffle(acc_indices)
if args.random_frames > len(vr):
    warnings.warn('Number of random frames reduced to {} to fit test video'.format(len(vr)))
    args.random_frames = len(vr)

parser = argparse.ArgumentParser("Decord benchmark")
parser.add_argument('--gpu', type=int, default=-1, help='context to run, use --gpu=-1 to use cpu only')
parser.add_argument('--file', type=str, default='/tmp/testsrc_h264_100s_default.mp4', help='Test video')
parser.add_argument('--seed', type=int, default=666, help='numpy random seed for random access indices')
parser.add_argument('--random-frames', type=int, default=300, help='number of random frames to run')
parser.add_argument('--width', type=int, default=320, help='resize frame width')
parser.add_argument('--height', type=int, default=240, help='resize frame height')

args = parser.parse_args()

test_video = args.file
if args.gpu > -1:
    ctx = de.gpu(args.gpu)
else:
    ctx = de.cpu()

vr = de.VideoReader(test_video, ctx, width=args.width, height=args.height)
cnt = 0
tic = time.time()
while True:
    try:
        frame = vr.next()
    except StopIteration:
        break
    cnt += 1
print(cnt, ' frames, elapsed time for sequential read: ', time.time() - tic)

np.random.seed(args.seed)  # fix seed for all random tests
acc_indices = np.arange(len(vr))
np.random.shuffle(acc_indices)
if args.random_frames > len(vr):

sys.path.insert(0, os.path.join(os.path.dirname(__file__), '../../python'))
import decord as de

parser = argparse.ArgumentParser("Decord benchmark")
parser.add_argument('--gpu', type=int, default=-1, help='context to run, use --gpu=-1 to use cpu only')
parser.add_argument('--file', type=str, default='/tmp/testsrc_h264_100s_default.mp4', help='Test video')
parser.add_argument('--seed', type=int, default=666, help='numpy random seed for random access indices')
parser.add_argument('--random-frames', type=int, default=300, help='number of random frames to run')
parser.add_argument('--width', type=int, default=320, help='resize frame width')
parser.add_argument('--height', type=int, default=240, help='resize frame height')

args = parser.parse_args()

test_video = args.file
if args.gpu > -1:
    ctx = de.gpu(args.gpu)
else:
    ctx = de.cpu()

vr = de.VideoReader(test_video, ctx, width=args.width, height=args.height)
cnt = 0
tic = time.time()
while True:
    try:
        frame = vr.next()
    except StopIteration:
        break
    cnt += 1
print(cnt, ' frames, elapsed time for sequential read: ', time.time() - tic)

np.random.seed(args.seed)  # fix seed for all random tests
acc_indices = np.arange(len(vr))

#     pip install decord

########################################################################
# Usage
# -----
#
# We provide some usage cases here to get you started. For complete API, please refer to official documentation.

################################################################
# Suppose we want to read a video. Let's download the example video first.
from gluoncv import utils
url = 'https://github.com/bryanyzhu/tiny-ucf101/raw/master/abseiling_k400.mp4'
video_fname = utils.download(url)

from decord import VideoReader
vr = VideoReader(video_fname)

################################################################
# If we want to load the video in a specific dimension so that it can be fed into a CNN for processing,

vr = VideoReader(video_fname, width=320, height=256)

################################################################
# Now we have loaded the video, if we want to know how many frames are there in the video,

duration = len(vr)
print('The video contains %d frames' % duration)

################################################################
# If we want to access frame at index 10,

frame = vr[9]

def __getitem__(self, idx):
        """
        Return:
            clips (torch.tensor), label (int)
        """
        record = self.video_records[idx]
        video_reader = decord.VideoReader(
            "{}.{}".format(os.path.join(self.video_dir, record.path), self.video_ext),
            # TODO try to add `ctx=decord.ndarray.gpu(0) or .cuda(0)`
        )
        record._num_frames = len(video_reader)

        offsets = self._sample_indices(record)
        clips = np.array([self._get_frames(video_reader, o) for o in offsets])

        if self.num_segments == 1:
            # [T, H, W, C] -> [C, T, H, W]
            return self.transforms(torch.from_numpy(clips[0])), record.label
        else:
            # [S, T, H, W, C] -> [S, C, T, H, W]
            return (
                torch.stack([
                    self.transforms(torch.from_numpy(c)) for c in clips

def __getitem__(self, idx):
        record = self.video_infos[idx]
        if self.use_decord:
            video_reader = decord.VideoReader('{}.{}'.format(
                osp.join(self.img_prefix, record.path), self.video_ext))
            record.num_frames = len(video_reader)
        else:
            video_reader = mmcv.VideoReader('{}.{}'.format(
                osp.join(self.img_prefix, record.path), self.video_ext))
            record.num_frames = len(video_reader)

        if self.test_mode:
            segment_indices, skip_offsets = self._get_test_indices(record)
        else:
            segment_indices, skip_offsets = self._sample_indices(
                record) if self.random_shift else self._get_val_indices(record)

        data = dict(
            num_modalities=DC(to_tensor(len(self.modalities))),
            gt_label=DC(to_tensor(record.label), stack=True, pad_dims=None))

#
# We provide some usage cases here to get you started. For complete API, please refer to official documentation.

################################################################
# Suppose we want to read a video. Let's download the example video first.
from gluoncv import utils
url = 'https://github.com/bryanyzhu/tiny-ucf101/raw/master/abseiling_k400.mp4'
video_fname = utils.download(url)

from decord import VideoReader
vr = VideoReader(video_fname)

################################################################
# If we want to load the video in a specific dimension so that it can be fed into a CNN for processing,

vr = VideoReader(video_fname, width=320, height=256)

################################################################
# Now we have loaded the video, if we want to know how many frames are there in the video,

duration = len(vr)
print('The video contains %d frames' % duration)

################################################################
# If we want to access frame at index 10,

frame = vr[9]
print(frame.shape)

################################################################
# For deep learning, usually we want to get multiple frames at once. Now you can use ``get_batch`` function,
# Suppose we want to get a 32-frame video clip by skipping one frame in between,

# Now we want to compare its speed with Opencv VideoCapture to demonstrate its efficiency.
# Let's load the same video and get all the frames randomly using both decoders to compare their performance.
# We will run the loading for 11 times: use the first one as warming up, and average the rest 10 runs as the average speed.

import cv2
import time
import numpy as np

frames_list = np.arange(duration)
np.random.shuffle(frames_list)

# Decord
for i in range(11):
    if i == 1:
        start_time = time.time()
    decord_vr = VideoReader(video_fname)
    frames = decord_vr.get_batch(frames_list)
end_time = time.time()
print('Decord takes %4.4f seconds.' % ((end_time - start_time)/10))

# OpenCV
for i in range(11):
    if i == 1:
        start_time = time.time()
    cv2_vr = cv2.VideoCapture(video_fname)
    for frame_idx in frames_list:
        cv2_vr.set(1, frame_idx)
        _, frame = cv2_vr.read()
    cv2_vr.release()
end_time = time.time()
print('OpenCV takes %4.4f seconds.' % ((end_time - start_time)/10))