vathos.data_loader

The Dataloader classes that can be used with vathos

Some DataVisualization

BG Images

FG_BG Images

FG_BG_MASK Images

DEPTH_FG_BG Images

Mean and Standard Deviation

bg_stat = (['0.573435604572296', '0.520844697952271', '0.457784473896027'], ['0.207058250904083', '0.208138316869736', '0.215291306376457'])
fg_bg_stat = (['0.568499565124512', '0.512103974819183', '0.452332496643066'], ['0.211068645119667', '0.211040720343590', '0.216081097722054'])
fg_bg_mask_stat = (['0.062296919524670', '0.062296919524670', '0.062296919524670'], ['0.227044790983200', '0.227044790983200', '0.227044790983200'])
depth_fg_bg_stat = (['0.302973538637161', '0.302973538637161', '0.302973538637161'], ['0.101284727454185', '0.101284727454185', '0.101284727454185'])

class DenseDepth(root, source_zipfolder, train=True, transform=None, target_transform=None)

DenseDepth Dataset

Parameters:

• root – the directory where to place the dataset unzipped files
• source_zipfiles – the directory where the dataset zip files are stored (mount your drive and give a absolute path)
• transform – torchvision transform for input images
• target_transform – torchvision transform for ouput images

Input is fg_bg image AND bg image

Target is fg_bg_mask AND depth_fg_bg image

static apply_on_batch(batch, apply_func)

applies a given function the batch

Parameters:

• batch – a batch of data
• apply_func – a function that is to be applied to each data in batch

extractall()

extracts the zip files into the root dir

static plot4_batch(batch)

Plots 4 images for batch

Parameters:batch – the batch of data of this DenseDepth dataset Returns:the created figure Return type:matplotlib.pyplot.figure

static plot_results(batch)

Plots 8 images for batch’s model results

Parameters:batch – the batch of data of this DenseDepth dataset Returns:the created figure Return type:matplotlib.pyplot.figure

static plot_sample(sample)

Plots a given sample of the dataset

Parameters:batch – the batch of data of this DenseDepth dataset Returns:the created figure Return type:matplotlib.pyplot.figure