Here are the classes, structs, unions and interfaces with brief descriptions:

[detail level 1234]

▼Ncaffe	A layer factory that allows one to register layers. During runtime, registered layers can be called by passing a LayerParameter protobuffer to the CreateLayer function:
►Ndb
CCursor
CDB
CTransaction
CAbsValLayer	Computes
CAccuracyLayer	Computes the classification accuracy for a one-of-many classification task
CAdaDeltaSolver
CAdaGradSolver
CAdamSolver	AdamSolver, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. Described in [1]
CArgMaxLayer	Compute the index of the max values for each datum across all dimensions $(C \times H \times W)$
CBaseConvolutionLayer	Abstract base class that factors out the BLAS code common to ConvolutionLayer and DeconvolutionLayer
CBaseDataLayer	Provides base for data layers that feed blobs to the Net
CBasePrefetchingDataLayer
CBatch
CBatchNormLayer	Normalizes the input to have 0-mean and/or unit (1) variance across the batch
CBatchReindexLayer	Index into the input blob along its first axis
CBiasLayer	Computes a sum of two input Blobs, with the shape of the latter Blob "broadcast" to match the shape of the former. Equivalent to tiling the latter Blob, then computing the elementwise sum
CBilinearFiller	Fills a Blob with coefficients for bilinear interpolation
CBlob	A wrapper around SyncedMemory holders serving as the basic computational unit through which Layers, Nets, and Solvers interact
►CBlockingQueue
Csync
CBNLLLayer	Computes $y = x + \log(1 + \exp(-x))$ if ; $y = \log(1 + \exp(x))$ otherwise
►CCaffe
►CRNG
CGenerator
CConcatLayer	Takes at least two Blobs and concatenates them along either the num or channel dimension, outputting the result
CConstantFiller	Fills a Blob with constant values
CContrastiveLossLayer	Computes the contrastive loss $E = \frac{1}{2N} \sum\limits_{n=1}^N \left(y\right) d^2 + \left(1-y\right) \max \left(margin-d, 0\right)^2$ where $d = \left\| \left\| a_n - b_n \right\| \right\|_2$ . This can be used to train siamese networks
CConvolutionLayer	Convolves the input image with a bank of learned filters, and (optionally) adds biases
CCPUTimer
CCropLayer	Takes a Blob and crop it, to the shape specified by the second input Blob, across all dimensions after the specified axis
CDataLayer
CDataTransformer	Applies common transformations to the input data, such as scaling, mirroring, substracting the image mean..
CDeconvolutionLayer	Convolve the input with a bank of learned filters, and (optionally) add biases, treating filters and convolution parameters in the opposite sense as ConvolutionLayer
CDropoutLayer	During training only, sets a random portion of to 0, adjusting the rest of the vector magnitude accordingly
CDummyDataLayer	Provides data to the Net generated by a Filler
CEltwiseLayer	Compute elementwise operations, such as product and sum, along multiple input Blobs
CELULayer	Exponential Linear Unit non-linearity $y = \left\{ \begin{array}{lr} x & \mathrm{if} \; x > 0 \\ \alpha (\exp(x)-1) & \mathrm{if} \; x \le 0 \end{array} \right.$
CEmbedLayer	A layer for learning "embeddings" of one-hot vector input. Equivalent to an InnerProductLayer with one-hot vectors as input, but for efficiency the input is the "hot" index of each column itself
CEuclideanLossLayer	Computes the Euclidean (L2) loss $E = \frac{1}{2N} \sum\limits_{n=1}^N \left\| \left\| \hat{y}_n - y_n \right\| \right\|_2^2$ for real-valued regression tasks
CExpLayer	Computes $y = \gamma ^ {\alpha x + \beta}$ , as specified by the scale $\alpha$ , shift $\beta$ , and base $\gamma$
CFiller	Fills a Blob with constant or randomly-generated data
CFilterLayer	Takes two+ Blobs, interprets last Blob as a selector and filter remaining Blobs accordingly with selector data (0 means that the corresponding item has to be filtered, non-zero means that corresponding item needs to stay)
CFlattenLayer	Reshapes the input Blob into flat vectors
CGaussianFiller	Fills a Blob with Gaussian-distributed values
CHDF5DataLayer	Provides data to the Net from HDF5 files
CHDF5OutputLayer	Write blobs to disk as HDF5 files
CHingeLossLayer	Computes the hinge loss for a one-of-many classification task
CIm2colLayer	A helper for image operations that rearranges image regions into column vectors. Used by ConvolutionLayer to perform convolution by matrix multiplication
CImageDataLayer	Provides data to the Net from image files
CInfogainLossLayer	A generalization of MultinomialLogisticLossLayer that takes an "information gain" (infogain) matrix specifying the "value" of all label pairs
CInnerProductLayer	Also known as a "fully-connected" layer, computes an inner product with a set of learned weights, and (optionally) adds biases
CInputLayer	Provides data to the Net by assigning tops directly
CInternalThread
CLayer	An interface for the units of computation which can be composed into a Net
CLayerRegisterer
CLayerRegistry
CLogLayer	Computes $y = log_{\gamma}(\alpha x + \beta)$ , as specified by the scale $\alpha$ , shift $\beta$ , and base $\gamma$
CLossLayer	An interface for Layers that take two Blobs as input – usually (1) predictions and (2) ground-truth labels – and output a singleton Blob representing the loss
CLRNLayer	Normalize the input in a local region across or within feature maps
CLSTMLayer	Processes sequential inputs using a "Long Short-Term Memory" (LSTM) [1] style recurrent neural network (RNN). Implemented by unrolling the LSTM computation through time
CLSTMUnitLayer	A helper for LSTMLayer: computes a single timestep of the non-linearity of the LSTM, producing the updated cell and hidden states
CMemoryDataLayer	Provides data to the Net from memory
CMSRAFiller	Fills a Blob with values $x \sim N(0, \sigma^2)$ where $\sigma^2$ is set inversely proportional to number of incoming nodes, outgoing nodes, or their average
CMultinomialLogisticLossLayer	Computes the multinomial logistic loss for a one-of-many classification task, directly taking a predicted probability distribution as input
CMVNLayer	Normalizes the input to have 0-mean and/or unit (1) variance
CNesterovSolver
►CNet	Connects Layers together into a directed acyclic graph (DAG) specified by a NetParameter
CCallback
CNeuronLayer	An interface for layers that take one blob as input ( ) and produce one equally-sized blob as output ( ), where each element of the output depends only on the corresponding input element
CParameterLayer
CPoolingLayer	Pools the input image by taking the max, average, etc. within regions
CPositiveUnitballFiller	Fills a Blob with values $x \in [0, 1]$ such that $\forall i \sum_j x_{ij} = 1$
CPowerLayer	Computes $y = (\alpha x + \beta) ^ \gamma$ , as specified by the scale $\alpha$ , shift $\beta$ , and power $\gamma$
CPReLULayer	Parameterized Rectified Linear Unit non-linearity $y_i = \max(0, x_i) + a_i \min(0, x_i)$ . The differences from ReLULayer are 1) negative slopes are learnable though backprop and 2) negative slopes can vary across channels. The number of axes of input blob should be greater than or equal to 2. The 1st axis (0-based) is seen as channels
CPythonLayer
CRecurrentLayer	An abstract class for implementing recurrent behavior inside of an unrolled network. This Layer type cannot be instantiated – instead, you should use one of its implementations which defines the recurrent architecture, such as RNNLayer or LSTMLayer
CReductionLayer	Compute "reductions" – operations that return a scalar output Blob for an input Blob of arbitrary size, such as the sum, absolute sum, and sum of squares
CReLULayer	Rectified Linear Unit non-linearity $y = \max(0, x)$ . The simple max is fast to compute, and the function does not saturate
CReshapeLayer
CRMSPropSolver
CRNNLayer	Processes time-varying inputs using a simple recurrent neural network (RNN). Implemented as a network unrolling the RNN computation in time
CScaleLayer	Computes the elementwise product of two input Blobs, with the shape of the latter Blob "broadcast" to match the shape of the former. Equivalent to tiling the latter Blob, then computing the elementwise product. Note: for efficiency and convenience, this layer can additionally perform a "broadcast" sum too when `bias_term: true` is set
CSGDSolver	Optimizes the parameters of a Net using stochastic gradient descent (SGD) with momentum
CSigmoidCrossEntropyLossLayer	Computes the cross-entropy (logistic) loss $E = \frac{-1}{n} \sum\limits_{n=1}^N \left[ p_n \log \hat{p}_n + (1 - p_n) \log(1 - \hat{p}_n) \right]$ , often used for predicting targets interpreted as probabilities
CSigmoidLayer	Sigmoid function non-linearity $y = (1 + \exp(-x))^{-1}$ , a classic choice in neural networks
CSignalHandler
CSilenceLayer	Ignores bottom blobs while producing no top blobs. (This is useful to suppress outputs during testing.)
CSliceLayer	Takes a Blob and slices it along either the num or channel dimension, outputting multiple sliced Blob results
CSoftmaxLayer	Computes the softmax function
CSoftmaxWithLossLayer	Computes the multinomial logistic loss for a one-of-many classification task, passing real-valued predictions through a softmax to get a probability distribution over classes
►CSolver	An interface for classes that perform optimization on Nets
CCallback
CSolverRegisterer
CSolverRegistry
CSplitLayer	Creates a "split" path in the network by copying the bottom Blob into multiple top Blobs to be used by multiple consuming layers
CSPPLayer	Does spatial pyramid pooling on the input image by taking the max, average, etc. within regions so that the result vector of different sized images are of the same size
CSyncedMemory	Manages memory allocation and synchronization between the host (CPU) and device (GPU)
CTanHLayer	TanH hyperbolic tangent non-linearity $y = \frac{\exp(2x) - 1}{\exp(2x) + 1}$ , popular in auto-encoders
CThresholdLayer	Tests whether the input exceeds a threshold: outputs 1 for inputs above threshold; 0 otherwise
CTileLayer	Copy a Blob along specified dimensions
CTimer
CUniformFiller	Fills a Blob with uniformly distributed values $x\sim U(a, b)$
CWindowDataLayer	Provides data to the Net from windows of images files, specified by a window data file. This layer is DEPRECATED and only kept for archival purposes for use by the original R-CNN
CXavierFiller	Fills a Blob with values $x \sim U(-a, +a)$ where is set inversely proportional to number of incoming nodes, outgoing nodes, or their average