Task1 食物声音识别-Baseline

Task1 食物声音识别-Baseline

!wget http://tianchi-competition.oss-cn-hangzhou.aliyuncs.com/531887/train_sample.zip

--2021-04-12 23:39:55--  http://tianchi-competition.oss-cn-hangzhou.aliyuncs.com/531887/train_sample.zip
Resolving tianchi-competition.oss-cn-hangzhou.aliyuncs.com (tianchi-competition.oss-cn-hangzhou.aliyuncs.com)... 118.31.232.194
Connecting to tianchi-competition.oss-cn-hangzhou.aliyuncs.com (tianchi-competition.oss-cn-hangzhou.aliyuncs.com)|118.31.232.194|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 540689175 (516M) [application/zip]
Saving to: ‘train_sample.zip’

100%[======================================>] 540,689,175 12.7MB/s   in 41s    

2021-04-12 23:40:36 (12.5 MB/s) - ‘train_sample.zip’ saved [540689175/540689175]

!unzip -qq train_sample.zip
!\rm train_sample.zip

!wget http://tianchi-competition.oss-cn-hangzhou.aliyuncs.com/531887/test_a.zip

--2021-04-12 23:41:46--  http://tianchi-competition.oss-cn-hangzhou.aliyuncs.com/531887/test_a.zip
Resolving tianchi-competition.oss-cn-hangzhou.aliyuncs.com (tianchi-competition.oss-cn-hangzhou.aliyuncs.com)... 118.31.232.194
Connecting to tianchi-competition.oss-cn-hangzhou.aliyuncs.com (tianchi-competition.oss-cn-hangzhou.aliyuncs.com)|118.31.232.194|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1092637852 (1.0G) [application/zip]
Saving to: ‘test_a.zip’

100%[====================================>] 1,092,637,852 13.9MB/s   in 85s    

2021-04-12 23:43:12 (12.2 MB/s) - ‘test_a.zip’ saved [1092637852/1092637852]

!unzip -qq test_a.zip
!\rm test_a.zip

# 基本库

import pandas as pd
import numpy as np

from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report
from sklearn.model_selection import GridSearchCV

from sklearn.preprocessing import MinMaxScaler

# 搭建分类模型所需要的库

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, Flatten, Dense, MaxPool2D, Dropout
from tensorflow.keras.utils import to_categorical 

from sklearn.ensemble import RandomForestClassifier
from sklearn.svm import SVC

!pip install librosa --user

# 其他库

import os
import librosa
import librosa.display
import glob 

!pip install tqdm

from tqdm import tqdm#可扩展的Python进度条，可以在 Python 长循环中添加一个进度提示信息，用户只需要封装任意的迭代器 tqdm(iterator)。
def extract_features(parent_dir, sub_dirs, max_file=10, file_text="*.wav"):#wav是最常见的声音文件格式之一，是微软公司专门为Windows开发的一种标准数字音频文件
    c =0
    label, feature = [],[]
    for sub_dir in sub_dirs:
        for fn in tqdm(glob.glob(os.path.join(parent_dir, sub_dir, file_text))[:max_file]):#os.path.join()函数：连接两个或更多的路径名组件Path20 = os.path.join(Path1,Path2,Path3)Path20 = home\develop\code
            label_name = fn.split('/')[-2]
            '''str = "Line1-abcdef \nLine2-abc \nLine4-abcd";
print str.split( );       # 以空格为分隔符，包含 \n
print str.split(' ', 1 ); # 以空格为分隔符，分隔成两个
['Line1-abcdef', 'Line2-abc', 'Line4-abcd']
['Line1-abcdef', '\nLine2-abc \nLine4-abcd']'''
            #append() 方法用于在列表末尾添加新的对象.extend() 函数用于在列表末尾一次性追加另一个序列中的多个值（用新列表扩展原来的列表）
            label.extend(label_dict[label_name])
            X, sample_rate = librosa.load(fn, res_type='kaiser_fast')#
            mels = np.mean(librosa.feature.melspectrogram(y=X, sr=sample_rate).T, axis=0)
            feature.extend([mels])
    return [feature, label]

parent_dir = './train_sample'
save_dir = './'
folds = sub_dirs = np.array(['aloe','burger','cabbage','candied_fruits',
                             'carrots','chips','chocolate','drinks','fries',
                            'grapes','gummies','ice-cream','jelly','noodles','pickles',
                            'pizza','ribs','salmon','soup','wings'])
# 获取特征feature以及类别的label
temp = extract_features(parent_dir,sub_dirs,max_file=100)

temp = np.array(temp)
data = temp.transpose()
'''根据维度进行转置'''

X = np.vstack(data[:, 0])
Y = np.array(data[:, 1])

X,shape
Y,shape

X_train, X_test, Y_train, Y_test = train_test_split(X, Y, random_state = 1, stratify=Y)
print('训练集的大小',len(X_train))
print('测试集的大小',len(X_test))

X_train = X_train.reshape(-1, 16, 8, 1)
X_test = X_test.reshape(-1, 16, 8, 1)

model = Sequential()

# 输入的大小
input_dim = (16, 8, 1)

model.add(Conv2D(64, (3, 3), padding = "same", activation = "tanh", input_shape = input_dim))# 卷积层
model.add(MaxPool2D(pool_size=(2, 2)))# 最大池化
model.add(Conv2D(128, (3, 3), padding = "same", activation = "tanh")) #卷积层
model.add(MaxPool2D(pool_size=(2, 2))) # 最大池化层
model.add(Dropout(0.1))
model.add(Flatten()) # 展开
model.add(Dense(1024, activation = "tanh"))
model.add(Dense(20, activation = "softmax")) # 输出层：20个units输出20个类的概率

# 编译模型，设置损失函数，优化方法以及评价标准
model.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = ['accuracy'])

model.summary()

# 训练模型
model.fit(X_train, Y_train, epochs = 20, batch_size = 15, validation_data = (X_test, Y_test))
'''validation_data用来在每个epoch之后，或者每几个epoch，验证一次验证集，用来及早发现问题，比如过拟合，或者超参数设置有问题。

这样可以方便我们及时调整参数

 

针对超参的选择我们是根据验证集上的效果来进行调整的，因此验证集可以看做参与到“人工调参”的训练过程；
2）注意训练集、验证集和测试集应该服从同一数据分布，这样我们才能进行玄学调参；
3）测试集是可以没有的，但验证集是必须有的，如果验证集具有足够的泛化代表性，是可以不需要设置测试集的；注意测试集的存在只是为了验证我们在训练集和验证集上进行模型的超参和参数训练后，验证我们得到的模型是否具有泛化性能！'''

def extract_features(test_dir, file_ext="*.wav"):
    feature = []
    for fn in tqdm(glob.glob(os.path.join(test_dir, file_ext))[:]): # 遍历数据集的所有文件
        X, sample_rate = librosa.load(fn,res_type='kaiser_fast')
        mels = np.mean(librosa.feature.melspectrogram(y=X,sr=sample_rate).T,axis=0) # 计算梅尔频谱(mel spectrogram),并把它作为特征
        feature.extend([mels])
    return feature

X_test = extract_features('./test_a/')

X_test = np.vstack(X_test)
predictions = model.predict(X_test.reshape(-1, 16, 8, 1))

preds = np.argmax(predictions, axis = 1)
preds = [label_dict_inv[x] for x in preds]

path = glob.glob('./test_a/*.wav')
result = pd.DataFrame({'name':path, 'label': preds})

result['name'] = result['name'].apply(lambda x: x.split('/')[-1])
result.to_csv('submit.csv',index=None)

!ls ./test_a/*.wav | wc -l

!wc -l submit.csv