制作大图效果

最新推荐文章于 2021-07-01 13:48:36 发布

所思即所得

最新推荐文章于 2021-07-01 13:48:36 发布

阅读量176

点赞数

分类专栏：云图分类

本文链接：https://blog.csdn.net/shouhan6396/article/details/89884771

版权

云图分类专栏收录该内容

5 篇文章 0 订阅

订阅专栏

模型预测

import pickle
import cv2
import numpy as np
f = open('/home/xm/桌面/论文精华/sample_data_498_558.txt','rb')
data = pickle.load(f)
pred = model.predict(data)

将分类结果以像素点的形式拼接

B = []
for i in range(len(pred)):
    B.append(abs(abs(pred[i][1])-abs(pred[i][2])))
BB = np.array(B)
BB = np.reshape(BB,(498,558))
data = np.argmax(pred,axis=1)
print(data.shape)
img = data.reshape([498,558])
pic = np.zeros([498,558,3])
for i in range(498):
    for j in range(558):
        if img[i][j] == 0:
            pic[i][j][0] = 0
            pic[i][j][1] = 0
            pic[i][j][2] = 0
        if img[i][j] == 1:
            pic[i][j][0] = 255
            pic[i][j][1] = 255
            pic[i][j][2] = 255
        if img[i][j] == 2:
            pic[i][j][0] = 0
            pic[i][j][1] = 0
            pic[i][j][2] = 255
        if (img[i][j] == 1 and BB[i][j] < 0.2) or (img[i][j] == 2 and BB[i][j] < 0.2):
            pic[i][j][0] = 255
            pic[i][j][1] = 0
            pic[i][j][2] = 0
cv2.imwrite('/home/xm/桌面/densenet_498*558_image.png',pic)