caffe2.python.layers.conv.Conv Class Reference

Inheritance diagram for caffe2.python.layers.conv.Conv:

Public Member Functions
def	__init__ (self, model, input_record, output_dim, kernel_h, kernel_w, stride_h, stride_w, pad_b=None, pad_l=None, pad_r=None, pad_t=None, order='NHWC', kernel_init=None, bias_init=None, kernel_optim=None, bias_optim=None, name='conv', kwargs)
def	add_ops (self, net)

Public Attributes
	kernel_h
	kernel_w
	stride_h
	stride_w
	pad_t
	pad_b
	pad_r
	pad_l
	order
	kernel
	bias
	output_schema

Detailed Description

    Convolutional layer
    Input:
    - input_record: at least has the shape info of C (num_channels)
    - output_dim: number of convolutional filters
    - kernel_h, kernel_w: kernel size for h and w
    - stride_h, stride_w: stride for h and w
    - pad_b, pad_l, pad_r, pad_t: padding sizes, if stride == 1,
                                  'None' value will do auto padding
    - order: either 'NHWC' or 'NCHW'

Definition at line 15 of file conv.py.

---

The documentation for this class was generated from the following file:

• caffe2/python/layers/conv.py