Keras：ValueError：无法将大小为4096的数组重塑为形状1,64,64,3

问题描述

我正在使用cv2为网络营地拍照 我使用ASL数据集训练了一个模型，然后将要对来自Webcamp的图像进行分类。为此，我需要根据训练后的模型调整图像的大小。

它说我必须将Webcamp图像转换为形状为（1、64、64、3），但这给了我一个错误。

import cv2
import numpy as np
from keras.models import load_model
from keras.preprocessing import image
import imutils
from numpy import array

model = load_model('trained_model2.h5')
print('model loaded!')
out_label=['A','B','C','D','E','F','G','H','I','K','L','M','N','O','P','Q','R','S','T','U','V','W','X','Y']

pre = []
s = ''
cchar = [0,0]
c1 = ''    
aWeight = 0.5    
capture = cv2.VideoCapture(0)    
top,right,bottom,left = 170,150,425,450    
num_frames = 0    
flag = 0
flag1=  0

while capture.isOpened():
    ret,frame = capture.read()
    
    frame = imutils.resize(frame,width=700)

    frame = cv2.flip(frame,1)

    clone = frame.copy()

    (height,width) = frame.shape[:2]

    roi = frame[top:bottom,right:left]
    
    gray = cv2.cvtColor(roi,cv2.COLOR_BGR2GRAY)
    gray = cv2.GaussianBlur(gray,(7,7),0)
    
    img = gray
    
    IMG_SIZE = 64
    
    img = cv2.resize(img,(64,64))
    img = np.expand_dims(img,axis=0)
    data = img.reshape(1,IMG_SIZE,3)
    print("converted")
         
    data = data / 255
   
    model_out = model.predict([data][0])
    pnb = np.argmax(model_out)
    print(str(np.argmax(model_out)) + " " + str(out_label[pnb]))
    pre.append(out_label[pnb]) 

    cv2.putText(clone,'%s ' % (str(out_label[pnb])),(450,150),cv2.FONT_HERShey_PLAIN,5,(0,255,0))
    
    cv2.rectangle(frame,(100,100),(300,300),0),0)
    crop_image = frame[100:300,100:300]
    
    blur = cv2.GaussianBlur(crop_image,(3,3),0)
    hsv = cv2.cvtColor(blur,cv2.COLOR_BGR2HSV)
    mask2 = cv2.inRange(hsv,np.array([2,0]),np.array([20,255]))
    kernel = np.ones((5,5))  
    dilation = cv2.dilate(mask2,kernel,iterations=1)
    erosion = cv2.erode(dilation,iterations=1)
    filtered = cv2.GaussianBlur(erosion,0)
    ret,thresh = cv2.threshold(filtered,127,0)
    
    cv2.imshow("Gesture",frame)
    cv2.imshow("Thresholded",thresh)
    
    if cv2.waitKey(1) == ord('q'):
        break
    
capture.release()
cv2.destroyAllWindows()

解决方法

您正在尝试将4096维图像重塑为形状为（64、64、3）的图像-这表示具有RGB颜色（或OpenCV中为BGR颜色）的图像。但是，正在读取的图像是灰度的。这意味着您不应将其重塑为（64，64，3），而应重塑为（64，64，1）。

data = img.reshape(1,IMG_SIZE,1)

keras opencv-python python