Creates two-branch residual block Parameters: ---------- incoming_layer : instance of Lasagne layer Parent layer ratio_n_fil

Creates stacked Lasagne layers ConvLayer -> BN -> (ReLu) Parameters: ---------- incoming_layer : instance of Lasagne layer Parent

This function produces a semi-redundant batch of training samples from the location 'p' in the provided string (data). For instance, assuming

Returns a string representation of the entire network contained under this layer. Parameters ---------- layer : Layer or lis

yields a random crop of size crop_size

Returns adjacency lists for each layer (node) in network. Warning: Assumes repr is unique to a layer instance, else this entire approach WILL

Pretty print a table provided as a list of lists.

Insert the header (first two lines) in the representation.

Data augmentation batch iterator for feeding images into CNN. Pads each image with 4 pixels on every side. Randomly crops image with orig

Batch iterator for fine tuning network, no augmentation.

yields center crop of size output_size (may be 1d or 2d) from data and seg

Downloads and extracts the dataset, if needed.

Sequential network, no branches or cycles

Two branches

Runs a generator in a background thread, caching up to `num_cached` items.

Loads the parameters for the caffe reference model. Also checks to make sure the two models produce equivalent results. The ouput model i

Converts an RGB image to caffe format and downscales it as needed by C3D Parameters ---------- im numpy array an RGB image

This function uses the user-provided string "generation_phrase" and current state of the RNN generate text. The function works in thr

Adapted from https://github.com/Lasagne/Recipes/tree/master/papers/deep_residual_learning. Tweaked to be consistent with 'Identity Mappings i

Adapted from https://github.com/Lasagne/Recipes/tree/master/papers/deep_residual_learning. Tweaked to be consistent with 'Identity Mappings i

Adapted from https://github.com/Lasagne/Recipes/tree/master/papers/deep_residual_learning. Tweaked to be consistent with 'Identity Mappings

Randomly augments images by horizontal flipping with a probability of `flip` and random translation of up to `trans` pixels in both direction

just a simple batch iterator, no cropping, no rotation, no anything

Creates a DenseNet model in Lasagne. Parameters ---------- input_shape : tuple The shape of the input layer, as ``(batchsize

Creates a DenseNet model in Lasagne. Parameters ---------- input_shape : tuple The shape of the input layer, as ``(batchsize

Builds C3D model Returns ------- dict A dictionary containing the network layers, where the output layer is at key 'prob'

Converts a Caffe format image back to the standard format, so that it can be plotted. Parameters ---------- raw_im numpy array

Converts a list of images to a 5d tensor that serves as input to C3D Parameters ---------- images: 4d numpy array or list of 3d numpy

Generates one epoch of batches of inputs and targets, optionally shuffled.

Returns a generator over `items`, printing the number and percentage of items processed and the estimated remaining processing time before yi

Sets the parameters of the model using the weights stored in model_file Parameters ---------- net: a Lasagne layer model_file: s