我正在开展一个项目,目标是检测性别并对图像进行分类。我做了一些研究,发现了一篇研究论文:Gil Levi 和 Tal Hassnar 的 AgeandGender Classificationusing Convolutional Neural Networks。我尝试在 Keras 中复制他们最初在 Caffe 中创建的深度网络。但问题是模型的准确率停留在 50%(基本上是随机抛硬币)。我做错了什么。任何帮助深表感谢。 顺便说一句,我使用 adience 数据集作为原始论文。 PS:我已经完全删除了 LRN 层,因为它们在 Keras 中不可用。 (我认为他们的缺席不应该损害模型的准确性) 这是代码。
#imports
import os
import numpy as np
from PIL import Image
import pickle
from keras.models import Sequential
from keras.callbacks import ModelCheckpoint
from keras.layers import Dense , Conv2D , Flatten , MaxPooling2D , Dropout , AveragePooling2D
from keras import initializers
from keras import optimizers
# creating the model object
gender_model = Sequential()
# adding layers to the model
# first convolutional layer
gender_model.add( Conv2D(96 , kernel_size=(7,7) , activation='relu' , strides=4 , input_shape=(227,227,3),
kernel_initializer= initializers.random_normal(stddev=0.01), use_bias = 1,
bias_initializer = 'Zeros' , data_format='channels_last'))
gender_model.add( MaxPooling2D(pool_size=3 , strides=2) )
gender_model.add( Conv2D(256, kernel_size=(5,5) , activation='relu', strides=1 , padding='same' , input_shape=(27,27,96),
kernel_initializer= initializers.random_normal(stddev=0.01) , use_bias=1,
bias_initializer='Ones' , data_format='channels_last') )
gender_model.add( MaxPooling2D(pool_size=3 , strides=2) )
# third convolutional layer
gender_model.add( Conv2D(384,kernel_size=(3,3) , activation='relu', strides=1 ,padding='same', input_shape=(13,13,256),
kernel_initializer= initializers.random_normal(stddev=0.01), use_bias=1,
bias_initializer = 'Zeros' , data_format='channels_last') )
gender_model.add( MaxPooling2D(pool_size=3 , strides=2) )
# Now we flatten the output of last convolutional layer
gender_model.add( Flatten() )
# Now we connect the fully connected layers
gender_model.add( Dense(512, activation='relu' , use_bias=1, kernel_initializer=initializers.random_normal(stddev=0.005),
bias_initializer='Ones') )
gender_model.add( Dropout(0.5))
# connecting another fully connected layer
gender_model.add( Dense(512 , activation='relu' , use_bias=1, kernel_initializer=initializers.random_normal(stddev=0.005),
bias_initializer='Ones'))
gender_model.add( Dropout(0.5))
# connecting the final layer
gender_model.add( Dense(2, activation='softmax' , use_bias=1, kernel_initializer=initializers.random_normal(stddev=0.01),
bias_initializer='Zeros'))
# compiling the model
sgd_optimizer = optimizers.SGD(lr= 0.0001 , decay=1e-7 , momentum=0.0, nesterov=False)
gender_model.compile(optimizer=sgd_optimizer , loss= 'categorical_crossentropy' , metrics=['accuracy'])
gender_model.summary()
# partioning the loaded data
X = np.load('/content/drive/My Drive/X.npy')
y = np.load('/content/drive/My Drive/y_m.npy')
X_train = X[:15000]
y_train = y[:15000]
X_val = X[15000:]
y_val = y[15000:]
## creating chkpt path
chkpt_path = 'weights-improvement-{epoch:02d}--{val_acc:.2f}.hdf5'
checkpoint = ModelCheckpoint(chkpt_path , monitor='val_acc' , verbose=1 , save_best_only=True , mode='max')
callback_list = [checkpoint]
#finally training the model
gender_model.fit(X_train, y_train,
batch_size=50,
epochs=100,
validation_data=(X_val , y_val),
shuffle=1,
callbacks = callback_list
)
Python 的 Deepface 包提供了开箱即用的年龄和性别预测功能。
!pip install deepface
from deepface import DeepFace
obj = DeepFace.analyze("img1.jpg", actions = ["age", "gender", "emotion", "race"])
print(obj["age"], " years old ", obj["gender"])
它还可以分析情感和种族。您可以删除不必要的操作。
我的方法有问题。我没有裁剪脸部。因此,该模型无法理解每张图像中的随机背景。
def process_image(image_path, gender_net):
# gender_list = ['Male', 'Female']
font = cv2.FONT_HERSHEY_SIMPLEX
# Initialize MTCNN face detection model
mtcnn = MTCNN(keep_all=True, device='cuda' if torch.cuda.is_available() else 'cpu')
image = cv2.imread(image_path)
# Convert the image to RGB
image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# Detect faces in the image
boxes, face_confidence = mtcnn.detect(image_rgb)
# frame = cv2.imread(image_path)
org_frame = image.copy()
male_count = 0
female_count = 0
face_count = 0
# Showing count
st.header("Photo Analysis")
# print ("Photo Analysis")
col1, col2 = st.columns([1,3])
data = {
'Track Id': [],
'faces': []
}
df = pd.DataFrame(data)
with col1:
st.write("Count of detected Faces")
table_placeholder = st.empty()
with col2:
st.write("Display Photo")
# print ("Photo place-holder")
photo_placeholder = st.empty()
for rect in boxes:
if face_confidence[face_count] > 0.9:
x1, y1, x2, y2 = map(int, rect)
# Given padding values (top, bottom, left, right)
# padding=50
padding_top = 12
padding_bottom = 12
padding_left = 12
padding_right = 12
# Calculate new bounding box coordinates with padding
new_x1 = x1 - padding_left
new_y1 = y1 - padding_top
new_x2 = x2 + padding_right
new_y2 = y2 + padding_bottom
# Calculate width and height of the bounding box
new_width = new_x2 - new_x1
new_height = new_y2 - new_y1
x = new_x1
y = new_y1
w = new_width
h = new_height
if y < 0:
continue
face_image = image_rgb[y:y+h, x:x+w].copy()
# Convert the face image to PIL Image
pil_image = Image.fromarray(face_image)
# Preprocess the face image
face_tensor = preprocess(pil_image)
face_tensor = face_tensor.unsqueeze(0).to(device) # Add batch dimension and move to device
# Set the threshold value (between 0 and 1)
threshold = 0.51
# Pass the preprocessed face image through the gender classification model
with torch.no_grad():
outputs = gender_net(face_tensor)
probabilities = nn.functional.softmax(outputs, dim=1)[0]
print ("probabilities - ",probabilities)
confidence, predicted = torch.max(probabilities, 0)
if confidence.item() > threshold:
predicted_gender = 'FeMale' if predicted.item() == 0 else 'male'
if predicted_gender == 'FeMale':
print ("Female")
female_count += 1
overlay_text = "%s" % (predicted_gender)
cv2.putText(org_frame, overlay_text, (x, y), font, 1, (255, 0, 0), 2, cv2.LINE_AA)
cv2.rectangle(org_frame, (x, y), (x+w, y+h), (255, 0, 0), 2)
else:
print ("Male")
male_count += 1
overlay_text = "%s" % (predicted_gender)
cv2.putText(org_frame, overlay_text, (x, y), font, 1, (0, 255, 0), 2, cv2.LINE_AA)
cv2.rectangle(org_frame, (x, y), (x+w, y+h), (0, 255, 0), 2)
else:
predicted_gender = 'Unknown'
face_count = face_count + 1
# # Convert frame to PIL Image and display
frame = cv2.cvtColor(org_frame, cv2.COLOR_BGR2RGB) # Convert from BGR to RGB color space
pil_image = Image.fromarray(frame)
photo_placeholder.image(pil_image, use_column_width=True)
# Update the table with male and female count
df = pd.DataFrame({'Track Id': ['Male', 'Female'], 'faces': [male_count, female_count]})
table_placeholder.write(df)
git-hub 链接到以下 - https://github.com/ndb796/Face-Gender-Classification-PyTorch