Caffe Study Series (£): Caffe source Analysis vector<blob<dtype>*>& Bottom

Transferred from: http://blog.csdn.net/qq_14975217/article/details/51524042

Blob:4 dimensions n x C x H x W;

Bottom[0], bottom[1] represents several inputs for the layer.

Bottom[0]->count (): Input, total number of dimensions (number of elements)

Bottom[0]->nums (): input, the number of blocks (block), the parameter also corresponds to Batch_size, that is, several pictures are entered at the same time
C: Is the number of convolution cores (filter), each convolution core produces a channel output, in the input layer, C is directly the image of the channel number;
There is also a variable, dim;: Dimension of each block (number of elements)

Visualization:
| xxxxx | xxxxx | xxxxx | xxxxx | xxxxx | xxxxx | xxxxx | xxxxx |

, Nums = 8, Dim = 5, Count = 5*8 = 40;

Caffe Study Series (£): Caffe source Analysis vector<blob<dtype>*>& Bottom (EXT)