手写卷积前向过程(无反向)

import numpy as np
def convv(feature, conv, stride):
    # (x+2p - kernel)//2 + 1
    conv_l = conv.shape[0]
    rows = feature.shape[0]
    cols = rows
    out_size = (rows - conv_l) // stride + 1
    outcome = np.zeros((out_size,out_size))
    i = j = 0
    m = n =0
    while i<rows:
        n = 0
    #for i in range(rows):
        while j < cols:
        #for j in range(cols):
            if j+conv_l <= cols and i + conv_l <= rows:
                print(feature[i:i+conv_l, j:j+conv_l].shape)
                outcome[m][n] = np.sum(feature[i:i+conv_l, j:j+conv_l] * conv)
                n += 1
                j += stride
        n = 0
        j = 0
        i += stride
        m += 1
    return outcome
feature = np.arange(81).reshape((9, 9))
conv = np.ones(9).reshape((3, 3))
stride = 3
a = convv(feature,conv,stride)
print(a)

附一张numpy的写法
在这里插入图片描述

评论 2
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值