python - 如何从相机调整阵列大小以适合人脸识别模型？

Question

好吧，长话短说，我做了一个人脸识别模型。现在，当我想从我的模型中进行预测时，我收到以下错误：

ValueError：检查输入时出错：预期 conv2d_3_input 有 4 个维度，但得到了形状为 (182、182、3) 的数组

我使用我的相机和面部检测工作并绘制了一个矩形，只是我的面部识别预测出错了。

我是机器学习的新手，我真的很清楚自己需要做什么，所以不要因为我的错误而评判我：/

我已经尝试了一些对某些人有用的解决方案，比如重塑等，但对我来说没有。

编辑：好的，到目前为止你们设法帮助我摆脱了数组错误，但是通过调整我的 roi_color 我收到新错误：OpenCV(4.1.1) C:\projects\opencv-python\opencv\modules\imgproc\src \resize.cpp:3363: 错误: (-215:Assertion failed) !dsize.empty() in function 'cv::hal::resize'

谢谢你的帮助，简

import numpy as np
import cv2
import random as rd
from pathlib import Path
import string
from keras.models import model_from_json


def Recog_face(x, y, roi_color, frame, recognizer_face):
    predictions = recognizer_face.predict(roi_color)
    color = (255, 255, 255) 
    stroke = 2   
    font = cv2.FONT_HERSHEY_SIMPLEX 
    return cv2.putText(frame, predictions, (x,y), font, 1, color, stroke, cv2.LINE_AA)

def Detect_face(x, y, frame):
    color = (255, 0, 0)
    stroke = 1 
    end_cord_x = x + w 
    end_cord_y = y + h 
    return cv2.rectangle(frame, (x, y), (end_cord_x, end_cord_y), color, stroke) 

face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')




recognizer_face = model_from_json(open("face_rec.json", "r").read()) 


recognizer_face.load_weights('face_rec.h5')



cap = cv2.VideoCapture(0)


while True:

    ret, frame = cap.read() 
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) 
    faces = face_cascade.detectMultiScale(gray, scaleFactor=1.5, 
                                         minNeighbors=5)

    for (x, y, w, h) in faces: 

        roi_gray = gray[y:y+h, x:x+w] 
        roi_color = frame[y:y+h, x:x+w]

        Detect_face(x, y, frame)

        # Face recognizer
        Recog_face(x, y, roi_color, frame, recognizer_face)





    cv2.imshow('frame', frame)
    if cv2.waitKey(20) & 0xFF == ord('q'):
        break 

cap.release()
cv2.destroyAllWindows()

from keras.models import Sequential
from keras.layers import Convolution2D
from keras.layers import MaxPooling2D
from keras.layers import Flatten
from keras.layers import Dense

classifier = Sequential()

# Convolution
classifier.add(Convolution2D(32, 3, 3, input_shape=(64, 64, 3), activation='relu')) 
# Maxpooling To reduce the number of nodes will get in flattening step
# Pooling
classifier.add(MaxPooling2D(pool_size = (2, 2))) 
# adding second Convolution layer
classifier.add(Convolution2D(32, 3, 3, activation='relu'))
classifier.add(MaxPooling2D(pool_size = (2, 2))) 
# Flatten
classifier.add(Flatten())
# Full connection
classifier.add(Dense(output_dim = 128, activation = 'relu')) # 128 hidden nodes in the hidden layer
classifier.add(Dense(output_dim = 5, activation = 'softmax')) # softmax function cause multiple outcome 
# Compiling the CNN
classifier.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = ['accuracy'])
print(classifier.summary()) 



from keras.preprocessing.image import ImageDataGenerator


train_datagen = ImageDataGenerator(rescale=1./255, 
                                  shear_range=0.2, 
                                  zoom_range=0.2,
                                  horizontal_flip=True)

test_datagen = ImageDataGenerator(rescale=1./255)


training_set = train_datagen.flow_from_directory('images/Training_set',
                                                target_size=(64, 64), 
                                                batch_size=32,
                                                color_mode='rgb',
                                                class_mode='categorical') 

test_set = test_datagen.flow_from_directory('images/Test_set',
                                            target_size=(64, 64),
                                            batch_size=32,
                                            color_mode='rgb',
                                            class_mode='categorical') 
classifier.fit_generator(training_set,
                        steps_per_epoch=380,
                        epochs=2,
                        validation_data= test_set,
                        validation_steps=127)


fer_json = classifier.to_json()
with open("face_rec.json", "w") as json_file:
    json_file.write(fer_json)
classifier.save_weights("face_rec.h5")  

标签： pythontensorflowkerasconv-neural-networkface-recognition