def test_random_crop_video(self):
numFrames = random.randint(4, 128)
height = random.randint(10, 32) * 2
width = random.randint(10, 32) * 2
oheight = random.randint(5, (height - 2) / 2) * 2
owidth = random.randint(5, (width - 2) / 2) * 2
clip = torch.randint(0,
256, (numFrames, height, width, 3),
dtype=torch.uint8)
result = Compose([
transforms.ToTensorVideo(),
transforms.RandomCropVideo((oheight, owidth)),
])(clip)
self.assertEqual(result.size(2), oheight)
self.assertEqual(result.size(3), owidth)
transforms.RandomCropVideo((oheight, owidth)).__repr__()
def __init__(self, size, consistent=True):
if isinstance(size, numbers.Number):
self.size = (int(size), int(size))
else:
self.size = size
self.consistent = consistent
self.operation_torch = transforms_video.RandomCropVideo(size)
def train_transform(s):
return transforms.Compose([
transforms_video.ToTensorVideo(),
transforms_video.RandomHorizontalFlipVideo(),
transforms_video.RandomResizeVideo((s, round(s * 1.5))),
transforms_video.NormalizeVideo(mean=[0.43216, 0.394666, 0.37645],
std=[0.22803, 0.22145, 0.216989]),
transforms_video.RandomCropVideo(s)
])
import torchvision.datasets as datasets
from torch.utils.data import DataLoader
import torchvision.transforms as transforms
import torchvision.transforms._transforms_video as v_transform
import torch
TRAIN_BATCH_SIZE = 128
TEST_BATCH_SIZE = 128
FRAME_LENGTH = 16
transform = transforms.Compose([
v_transform.ToTensorVideo(),
v_transform.NormalizeVideo(mean=[0.43216, 0.394666, 0.37645],
std=[0.22803, 0.22145, 0.216989]),
v_transform.RandomHorizontalFlipVideo(),
v_transform.RandomCropVideo(112),
])
def custom_collate(batch):
filtered_batch = []
for video, _, label in batch:
filtered_batch.append((video, label))
return torch.utils.data.dataloader.default_collate(filtered_batch)
trainset = datasets.UCF101(
root='data/UCF101/UCF-101',
annotation_path=
'data/UCF101TrainTestSplits-RecognitionTask/ucfTrainTestlist',
frames_per_clip=FRAME_LENGTH,