image_analysis.decode package¶

Submodules¶

image_analysis.decode.extract_patches.extract_patches(mpeg_path, patch_dims, save_dir, *, n_frames=None, batch_size=250000)[source]¶

DESCRIPTION:

extract patches (neighborhoods) of pixels from videos in batches and save them to disk

ARGS:

save_dir:	directory to save extracted patches
mpeg_path:	file name for an mpeg video with one color channel
patch_dims:	size of patches to extract
n_frames:	number of frames to extract
batch_size:	size limit for each batch

image_analysis.decode.extract_patches.get_mpeg_dims(vid_fname)[source]¶

ARGS:

vid_fname:	file name for an mpeg video with one color channel

RETURNS:

(tuple):	containing number of rows, columns, and frames of data

image_analysis.decode.video_decoder.decode_mpeg(v_path, batch_size=1, stride=1, start_idx=0, end_idx=-1, pad=True)[source]¶

DESCRIPTION:

Creates a list of batches of frames from an MPEG file.

ARGS:

v_path:	Path to MPEG video (i.e. include the video’s name & extension)
batch_size:	Number of frames in each batch
stride:	Stride indicates beginning of batches, i.e. every stride’th frame (integer > 1)
start_idx:	Index of first frame for first batch (integer >= 0)
end_idx:	Index of last frame in the range of interest (integer >= 0)
pad:	Boolean value indicating whether the last batch should be padded if it is not full after decoding the mpeg

RETURNS:

LIST of NUMPY batches of frames (length x width x channels), and if the last batch is not full, it is padded with frames of: np.zeros((frame.shape))

image_analysis.decode.video_decoder.pad_batch(batch, batch_size, frame, pad=True)[source]¶

DESCRIPTION:

Take in a batch, pad it with 0s if necessary and return appended batch

ARGS:

frame:	a numpy array extracted from an MPEG (length x width x channel)
batch:	list of frames
batch_size:	number of frames per batch (integer >= 1)

RETURNS:

batch: list of a batch, batch has batch_size frames, each an ndarray: (L x W x C)