def pack_all():
trainSet, validationSet = trainval.read(
'//msralab/ProjectData/ehealth02/v-dinliu/Flow2D/Data/FlyingChairs_release/FlyingChairs_train_val.txt'
)
n = 64
for name, Set in [('train', trainSet), ('val', validationSet)]:
bn = 0
for i in range(0, len(Set), n):
subset = Set[i:i + n]
trainImg1 = [
ppm.load(
'//msralab/ProjectData/ehealth02/v-dinliu/Flow2D/Data/FlyingChairs_release/data/'
+ ('%05d' % i) + '_img1.ppm') for i in subset
]
trainImg2 = [
ppm.load(
'//msralab/ProjectData/ehealth02/v-dinliu/Flow2D/Data/FlyingChairs_release/data/'
+ ('%05d' % i) + '_img2.ppm') for i in subset
]
trainFlow = [
flo.load(
'//msralab/ProjectData/ehealth02/v-dinliu/Flow2D/Data/FlyingChairs_release/data/'
+ ('%05d' % i) + '_flow.flo') for i in subset
]
prefix = r'\\msralab\ProjectData\ScratchSSD\Users\v-dinliu\data\FlyingChairsBlock'
pack_data(
os.path.join(prefix,
'{}{}_{}.bin'.format(name, bn, len(subset))),
trainImg1, trainImg2, trainFlow)
bn += 1
print('{}/{}'.format(i, len(Set)))
from pympler.asizeof import asizeof
trainImg1 = [cv2.imread(file).astype('uint8') for file in things3d_dataset['image_0'][:samples:args.shard]]
print(asizeof(trainImg1[0]))
print(asizeof(trainImg1))
trainImg2 = [cv2.imread(file).astype('uint8') for file in things3d_dataset['image_1'][:samples:args.shard]]
print(asizeof(trainImg2[0]))
print(asizeof(trainImg2))
trainFlow = [things3d.load(file).astype('float16') for file in things3d_dataset['flow'][:samples:args.shard]]
print(asizeof(trainFlow[0]))
print(asizeof(trainFlow))
trainSize = len(trainFlow)
training_datasets = [(trainImg1, trainImg2, trainFlow)] * batch_size
print(asizeof(training_datasets))
# validation- chairs
_, validationSet = trainval.read(chairs_split_file)
validationSet = validationSet[:samples]
validationImg1 = [ppm.load(os.path.join(chairs_path, '%05d_img1.ppm' % i)) for i in validationSet]
validationImg2 = [ppm.load(os.path.join(chairs_path, '%05d_img2.ppm' % i)) for i in validationSet]
validationFlow = [flo.load(os.path.join(chairs_path, '%05d_flow.flo' % i)) for i in validationSet]
validationSize = len(validationFlow)
validation_datasets['chairs'] = (validationImg1, validationImg2, validationFlow)
'''
# validation- sintel
sintel_dataset = sintel.list_data()
divs = ('training',) if not getattr(config.network, 'class').get() == 'MaskFlownet' else ('training2',)
for div in divs:
for k, dataset in sintel_dataset[div].items():
img1, img2, flow, mask = [[sintel.load(p) for p in data] for data in zip(*dataset)]
validationSize += len(flow)
# load training set and validation set
if args.debug or args.fake_data:
trainSet = np.arange(0, 128)
validationSet = np.arange(0, 128)
trainImg1 = np.random.normal(size=(128, 384, 512, 3))
trainImg2 = np.random.normal(size=(128, 384, 512, 3))
trainFlow = np.random.normal(size=(128, 384, 512, 2))
validationImg1 = np.random.normal(size=(128, 384, 512, 3))
validationImg2 = np.random.normal(size=(128, 384, 512, 3))
validationFlow = np.random.normal(size=(128, 384, 512, 2))
trainSize = validationSize = 128
elif args.net_data:
print('socket data ...')
trainSet, validationSet = trainval.read(
'//msralab/ProjectData/ehealth02/v-dinliu/Flow2D/Data/FlyingChairs_release/FlyingChairs_train_val.txt'
)
client = DatasetClient()
trainSize = len(trainSet)
validationSize = len(validationSet)
validationImg1, validationImg2, validationFlow = zip(
*[fetch_data(i) for i in validationSet])
elif train_cfg.dataset.value == 'sintel':
print('loading sintel dataset ...')
subsets = train_cfg.subsets.value
print(subsets[0], subsets[1])
trainImg1 = []
trainImg2 = []
trainFlow = []
sintel_dataset = sintel.list_data(sintel.sintel_path)
