卷积函数演示

ゞ浴缸里的玫瑰 2022-03-09 10:31 189阅读 0赞

直接上代码吧：

import numpy as np
    import tensorflow as tf
    
    #此代码仅用于演示conv2d的参数意义，主要是padding，stride
    #先以2维矩阵测试，通道为channels
    input_channels=1
    input_data=np.reshape(np.linspace(1.0,60.0*input_channels,num=60*input_channels),[1,6,10,input_channels])
    filter_data=np.random.randint(-1,1,[3,3,input_channels,1])
    if input_channels==1:
        squeeze_input=tf.squeeze(input_data,axis=[0,3])
        squeeze_filter=tf.squeeze(filter_data,axis=[2,3])
    else:
        squeeze_input=tf.squeeze(input_data,axis=[0])
        squeeze_filter=tf.squeeze(filter_data,axis=[3])
    
    conv_valid_result=tf.nn.conv2d(input=input_data,filter=filter_data,strides=[1,1,1,1],padding="VALID")
    conv_same_result=tf.nn.conv2d(input=input_data,filter=filter_data,strides=[1,1,1,1],padding="SAME")
    sess=tf.Session()
    print("----------------------------待卷积的数据---------------------------")
    print(sess.run(squeeze_input))
    print('----------------------------卷积核数据---------------------------')
    print(sess.run(squeeze_filter))
    cv1=sess.run(conv_valid_result)
    print('-------------------------VALID（窄卷积结果）---------------------------')
    if input_channels==1:
        print(sess.run(tf.squeeze(cv1,axis=[0,3])))
    else:
        print(sess.run(tf.squeeze(cv1,axis=[0])))
    
    cv2=sess.run(conv_same_result)
    print('-------------------------SAME（同卷积结果）---------------------------')
    if input_channels==1:
        print(sess.run(tf.squeeze(cv2,axis=[0,3])))
    else:
        print(sess.run(tf.squeeze(cv2,axis=[0])))

运算结果如下：  
\----------------------------待卷积的数据---------------------------  
\[\[ 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.\]  
\[11. 12. 13. 14. 15. 16. 17. 18. 19. 20.\]  
\[21. 22. 23. 24. 25. 26. 27. 28. 29. 30.\]  
\[31. 32. 33. 34. 35. 36. 37. 38. 39. 40.\]  
\[41. 42. 43. 44. 45. 46. 47. 48. 49. 50.\]  
\[51. 52. 53. 54. 55. 56. 57. 58. 59. 60.\]\]  
\----------------------------卷积核数据---------------------------  
\[\[-1 0 0\]  
\[ 0 0 0\]  
\[-1 -1 0\]\]  
\-------------------------VALID（窄卷积结果）---------------------------  
\[\[ -44. -47. -50. -53. -56. -59. -62. -65.\]  
\[ -74. -77. -80. -83. -86. -89. -92. -95.\]  
\[-104. -107. -110. -113. -116. -119. -122. -125.\]  
\[-134. -137. -140. -143. -146. -149. -152. -155.\]\]  
\-------------------------SAME（同卷积结果）---------------------------  
\[\[ -11. -23. -25. -27. -29. -31. -33. -35. -37. -39.\]  
\[ -21. -44. -47. -50. -53. -56. -59. -62. -65. -68.\]  
\[ -31. -74. -77. -80. -83. -86. -89. -92. -95. -98.\]  
\[ -41. -104. -107. -110. -113. -116. -119. -122. -125. -128.\]  
\[ -51. -134. -137. -140. -143. -146. -149. -152. -155. -158.\]  
\[ 0. -41. -42. -43. -44. -45. -46. -47. -48. -49.\]\]