狗狗识别-百度西安交通大学大数据比赛baseline=0.2代码

题目简单介绍

官方提供100类狗的图片，选手将狗进行分类。

分析

官方提供的图片如下：





从图片可以看出背景复杂，图片中不仅有狗还有人，可能还有其他物体的背景，因此第一步需要将狗提取出来。考虑到从先训练一个狗的检测器将耗费一定的时间，因此采用目标检测SSD方法进行狗的检测。将狗检测出来再进行分类。

因此流程就是 检测 + 分类。

狗的检测核心代码

该代码使用Qt编写。项目配置如下：

INCLUDEPATH += /home/young/deeplearning/SSD/caffe/include \
+= /usr/include \
+= /home/young/deeplearning/SSD/caffe/src \
+= /home/young/software/cuda/include

LIBS += -L/home/young/deeplearning/SSD/caffe/build/lib -lcaffe
LIBS += -L/usr/lib/x86_64-linux-gnu -lopencv_core -lopencv_imgproc -lopencv_highgui

LIBS += -lglog -lgflags -lprotobuf -lboost_system -lboost_thread  -latlas

SOURCES += \
main.cpp

# cuda
INCLUDEPATH += /usr/local/cuda/include

LIBS += -L/usr/local/cuda/lib64 -lcudart -lcublas -lcurand
#cudnn
LIBS += -L/home/young/software/cuda/lib64 -lcudnn

SSD检测代码

#include "caffe/caffe.hpp"
#define USE_OPENCV
#define CPU_ONLY

#ifdef USE_OPENCV
#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#endif  // USE_OPENCV
#include <algorithm>
#include <iomanip>
#include <iosfwd>
#include <memory>
#include <string>
#include <utility>
#include <vector>
#include<fstream>

#ifdef USE_OPENCV
using namespace caffe;  // NOLINT(build/namespaces)

std::vector<std::string> splitEx(const std::string& src, std::string separate_character)
{
std::vector<std::string> strs;

int separate_characterLen = separate_character.size();//分割字符串的长度,这样就可以支持如“,,”多字符串的分隔符
int lastPosition = 0,index = -1;
while (-1 != (index = src.find(separate_character,lastPosition)))
{
strs.push_back(src.substr(lastPosition,index - lastPosition));
lastPosition = index + separate_characterLen;
}
std::string lastString = src.substr(lastPosition);//截取最后一个分隔符后的内容
if (!lastString.empty())
strs.push_back(lastString);//如果最后一个分隔符后还有内容就入队
return strs;
}

class Detector {
public:
Detector(const string& model_file,
const string& weights_file,
const string& mean_file,
const string& mean_value);

std::vector<vector<float> > Detect(const cv::Mat& img);

private:
void SetMean(const string& mean_file, const string& mean_value);

void WrapInputLayer(std::vector<cv::Mat>* input_channels);

void Preprocess(const cv::Mat& img,
std::vector<cv::Mat>* input_channels);

private:
shared_ptr<Net<float> > net_;
cv::Size input_geometry_;
int num_channels_;
cv::Mat mean_;
};

Detector::Detector(const string& model_file,
const string& weights_file,
const string& mean_file,
const string& mean_value) {
#ifdef CPU_ONLY
Caffe::set_mode(Caffe::CPU);
#else
Caffe::set_mode(Caffe::GPU);
#endif

/* Load the network. */
net_.reset(new Net<float>(model_file, TEST));
net_->CopyTrainedLayersFrom(weights_file);

CHECK_EQ(net_->num_inputs(), 1) << "Network should have exactly one input.";
CHECK_EQ(net_->num_outputs(), 1) <
4000
< "Network should have exactly one output.";

Blob<float>* input_layer = net_->input_blobs()[0];
num_channels_ = input_layer->channels();
CHECK(num_channels_ == 3 || num_channels_ == 1)
<< "Input layer should have 1 or 3 channels.";
input_geometry_ = cv::Size(input_layer->width(), input_layer->height());

/* Load the binaryproto mean file. */
SetMean(mean_file, mean_value);
}

std::vector<vector<float> > Detector::Detect(const cv::Mat& img) {
Blob<float>* input_layer = net_->input_blobs()[0];
input_layer->Reshape(1, num_channels_,
input_geometry_.height, input_geometry_.width);
/* Forward dimension change to all layers. */
net_->Reshape();

std::vector<cv::Mat> input_channels;
WrapInputLayer(&input_channels);

Preprocess(img, &input_channels);

net_->Forward();

/* Copy the output layer to a std::vector */
Blob<float>* result_blob = net_->output_blobs()[0];
const float* result = result_blob->cpu_data();
const int num_det = result_blob->height();
vector<vector<float> > detections;
for (int k = 0; k < num_det; ++k) {
if (result[0] == -1) {
// Skip invalid detection.
result += 7;
continue;
}
vector<float> detection(result, result + 7);
detections.push_back(detection);
result += 7;
}
return detections;
}

/* Load the mean file in binaryproto format. */
void Detector::SetMean(const string& mean_file, const string& mean_value) {
cv::Scalar channel_mean;
if (!mean_file.empty()) {
CHECK(mean_value.empty()) <<
"Cannot specify mean_file and mean_value at the same time";
BlobProto blob_proto;
ReadProtoFromBinaryFileOrDie(mean_file.c_str(), &blob_proto);

/* Convert from BlobProto to Blob<float> */
Blob<float> mean_blob;
mean_blob.FromProto(blob_proto);
CHECK_EQ(mean_blob.channels(), num_channels_)
<< "Number of channels of mean file doesn't match input layer.";

/* The format of the mean file is planar 32-bit float BGR or grayscale. */
std::vector<cv::Mat> channels;
float* data = mean_blob.mutable_cpu_data();
for (int i = 0; i < num_channels_; ++i) {
/* Extract an individual channel. */
cv::Mat channel(mean_blob.height(), mean_blob.width(), CV_32FC1, data);
channels.push_back(channel);
data += mean_blob.height() * mean_blob.width();
}

/* Merge the separate channels into a single image. */
cv::Mat mean;
cv::merge(channels, mean);

/* Compute the global mean pixel value and create a mean image
* filled with this value. */
channel_mean = cv::mean(mean);
mean_ = cv::Mat(input_geometry_, mean.type(), channel_mean);
}
if (!mean_value.empty()) {
CHECK(mean_file.empty()) <<
"Cannot specify mean_file and mean_value at the same time";
stringstream ss(mean_value);
vector<float> values;
string item;
while (getline(ss, item, ',')) {
float value = std::atof(item.c_str());
values.push_back(value);
}
CHECK(values.size() == 1 || values.size() == num_channels_) <<
"Specify either 1 mean_value or as many as channels: " << num_channels_;

std::vector<cv::Mat> channels;
for (int i = 0; i < num_channels_; ++i) {
/* Extract an individual channel. */
cv::Mat channel(input_geometry_.height, input_geometry_.width, CV_32FC1,
cv::Scalar(values[i]));
channels.push_back(channel);
}
cv::merge(channels, mean_);
}
}

/* Wrap the input layer of the network in separate cv::Mat objects
* (one per channel). This way we save one memcpy operation and we
* don't need to rely on cudaMemcpy2D. The last preprocessing
* operation will write the separate channels directly to the input
* layer. */
void Detector::WrapInputLayer(std::vector<cv::Mat>* input_channels) {
Blob<float>* input_layer = net_->input_blobs()[0];

int width = input_layer->width();
int height = input_layer->height();
float* input_data = input_layer->mutable_cpu_data();
for (int i = 0; i < input_layer->channels(); ++i) {
cv::Mat channel(height, width, CV_32FC1, input_data);
input_channels->push_back(channel);
input_data += width * height;
}
}

void Detector::Preprocess(const cv::Mat& img,
std::vector<cv::Mat>* input_channels) {
/* Convert the input image to the input image format of the network. */
cv::Mat sample;
if (img.channels() == 3 && num_channels_ == 1)
cv::cvtColor(img, sample, cv::COLOR_BGR2GRAY);
else if (img.channels() == 4 && num_channels_ == 1)
cv::cvtColor(img, sample, cv::COLOR_BGRA2GRAY);
else if (img.channels() == 4 && num_channels_ == 3)
cv::cvtColor(img, sample, cv::COLOR_BGRA2BGR);
else if (img.channels() == 1 && num_channels_ == 3)
cv::cvtColor(img, sample, cv::COLOR_GRAY2BGR);
else
sample = img;

cv::Mat sample_resized;
if (sample.size() != input_geometry_)
cv::resize(sample, sample_resized, input_geometry_);
else
sample_resized = sample;

cv::Mat sample_float;
if (num_channels_ == 3)
sample_resized.convertTo(sample_float, CV_32FC3);
else
sample_resized.convertTo(sample_float, CV_32FC1);

cv::Mat sample_normalized;
cv::subtract(sample_float, mean_, sample_normalized);

/* This operation will write the separate BGR planes directly to the
* input layer of the network because it is wrapped by the cv::Mat
* objects in input_channels. */
cv::split(sample_normalized, *input_channels);

CHECK(reinterpret_cast<float*>(input_channels->at(0).data)
== net_->input_blobs()[0]->cpu_data())
<< "Input channels are not wrapping the input layer of the network.";
}

DEFINE_string(mean_file, "",
"The mean file used to subtract from the input image.");
DEFINE_string(mean_value, "104,117,123",
"If specified, can be one value or can be same as image channels"
" - would subtract from the corresponding channel). Separated by ','."
"Either mean_file or mean_value should be provided, not both.");
DEFINE_string(file_type, "image",
"The file type in the list_file. Currently support image and video.");
DEFINE_string(out_file, "",
"If provided, store the detection results in the out_file.");
DEFINE_double(confidence_threshold, 0.7,
"Only store detections with score higher than the threshold.");

int main(int argc, char** argv) {
::google::InitGoogleLogging(argv[0]);
// Print output to stderr (while still logging)
FLAGS_alsologtostderr = 1;

#ifndef GFLAGS_GFLAGS_H_
namespace gflags = google;
#endif

gflags::SetUsageMessage("Do detection using SSD mode.\n"
"Usage:\n"
"    ssd_detect [FLAGS] model_file weights_file list_file\n");
gflags::ParseCommandLineFlags(&argc, &argv, true);
/*
if (argc < 4) {
gflags::ShowUsageWithFlagsRestrict(argv[0], "examples/ssd/ssd_detect");
return 1;
}*/

const string& model_file = "./model/deploy.prototxt";
const string& weights_file = "./model/VGG_VOC0712Plus_SSD_300x300_iter_240000.caffemodel";
const string& mean_file = FLAGS_mean_file;
const string& mean_value = FLAGS_mean_value;
const string& file_type = FLAGS_file_type;
const string& out_file = FLAGS_out_file;
const float confidence_threshold = FLAGS_confidence_threshold;

// Initialize the network.
Detector detector(model_file, weights_file, mean_file, mean_value);

// Set the output mode.

std::streambuf* buf = std::cout.rdbuf();
std::ofstream outfile;
if (!out_file.empty()) {
outfile.open(out_file.c_str());
if (outfile.good()) {
buf = outfile.rdbuf();
}
}
std::ostream out(buf);

// Process image one by one.
std::ifstream infile("./img/val.txt");
std::string file;

std::ofstream outTrainfile;
//outTrainfile.open("./train.txt");

while (std::getline(infile, file)) {
outTrainfile.open("./val.txt", std::ios::app);

std::vector<std::string> line = splitEx(file, " ");
std::string imgPath = line[0];
std::string label = line[1].substr(0,line[1].find("\r"));

std::string prefixImg = splitEx(imgPath, ".")[0];
std::string postfixImg = splitEx(imgPath, ".")[1];

if (file_type == "image") {
cv::Mat img = cv::imread("./img/Img/" + imgPath, -1);
CHECK(!img.empty()) << "Unable to decode image " << file;
std::vector<vector<float> > detections = detector.Detect(img);

/* Print the detection results. */
int index = 0;
for (int i = 0; i < detections.size(); ++i) {
const vector<float>& d = detections[i];
// Detection format: [image_id, label, score, xmin, ymin, xmax, ymax].
CHECK_EQ(d.size(), 7);
const float score = d[2];
if (score >= confidence_threshold && static_cast<int>(d[1]) == 12) {

out << imgPath << " ";
out << label << " ";
out << static_cast<int>(d[1]) << " ";
out << score << " ";
out << static_cast<int>(d[3] * img.cols) << " ";
out << static_cast<int>(d[4] * img.rows) << " ";
out << static_cast<int>(d[5] * img.cols) << " ";
out << static_cast<int>(d[6] * img.rows) << std::endl;

int xmin = static_cast<int>(d[3] * img.cols);
int ymin = static_cast<int>(d[4] * img.rows);
int w  =  static_cast<int>(d[5] * img.cols);
int h = static_cast<int>(d[6] * img.rows);

if(xmin < 0) xmin = 0;
if(ymin < 0) ymin = 0;
if(xmin + w > img.cols) w = img.cols - xmin;
if(ymin + h > img.rows) h = img.rows - ymin;

cv::Mat saveImg(img, cv::Rect(xmin, ymin, w, h));

std::ostringstream stream;
stream<<index;
std::string newImgPath = prefixImg + "_" + stream.str() + "."+postfixImg;
//outTrainfile<< newImgPath << " " << label << "\n";
std::string savePath = "./val/" + label  + "/" + newImgPath;
outTrainfile << savePath << " " << label << "\n";
cv::imwrite(savePath, saveImg);
index++;
outTrainfile.close();
}
}
} else if (file_type == "video") {
cv::VideoCapture cap(file);
if (!cap.isOpened()) {
LOG(FATAL) << "Failed to open video: " << file;
}
cv::Mat img;
int frame_count = 0;
while (true) {
bool success = cap.read(img);
if (!success) {
LOG(INFO) << "Process " << frame_count << " frames from " << file;
break;
}
CHECK(!img.empty()) << "Error when read frame";
std::vector<vector<float> > detections = detector.Detect(img);

/* Print the detection results. */
for (int i = 0; i < detections.size(); ++i) {
const vector<float>& d = detections[i];
// Detection format: [image_id, label, score, xmin, ymin, xmax, ymax].
CHECK_EQ(d.size(), 7);
const float score = d[2];
if (score >= confidence_threshold) {
/*
out << file << "_";
out << std::setfill('0') << std::setw(6) << frame_count << " ";
out << static_cast<int>(d[1]) << " ";
out << score << " ";
out << static_cast<int>(d[3] * img.cols) << " ";
out << static_cast<int>(d[4] * img.rows) << " ";
out << static_cast<int>(d[5] * img.cols) << " ";
out << static_cast<int>(d[6] * img.rows) << std::endl;*/
}
}
++frame_count;
}
if (cap.isOpened()) {
cap.release();
}
} else {
LOG(FATAL) << "Unknown file_type: " << file_type;
}
}
outTrainfile.close();
return 0;
}
#else
int main(int argc, char** argv) {
LOG(FATAL) << "This example requires OpenCV; compile with USE_OPENCV.";
}
#endif  // USE_OPENCV

分类器选择

随着深度学习的发展，涌现出一些效果不错的网络，有AlexNet, VGG,ResNet,google系列的V1,V2,V3,V4。从ImageNet结果可以看出网络越深,分类效果越好。而且这些网络都公开了训练好的模型，由于比赛训练数据有限，采用迁移学习的思想进行网络学习。本人实验过VGG, ResNet18，V3，V3的网络效果最好。

训练代码如下：

from keras.applications.inception_v3 import InceptionV3
import os
from keras.layers import Flatten, Dense, AveragePooling2D
from keras.models import Model
from keras.optimizers import RMSprop, SGD
from keras.callbacks import ModelCheckpoint
from keras.preprocessing.image import ImageDataGenerator

learning_rate = 0.0001
img_width = 299
img_height = 299
nbr_train_samples = 3019
nbr_validation_samples = 758
nbr_epochs = 25
batch_size = 1
nb_classes= 100

train_data_dir = './train'
val_data_dir = './val'

DogNames = []
for i in range(0,100):
DogNames.append(str(i))

print('Loading InceptionV3 Weights ...')
InceptionV3_notop = InceptionV3(include_top=False, weights='imagenet',
input_tensor=None, input_shape=(299, 299, 3))
print('Adding Average Pooling Layer and Softmax Output Layer ...')
output = InceptionV3_notop.get_layer(index = -1).output  # Shape: (8, 8, 2048)
output = AveragePooling2D((8, 8), strides=(8, 8), name='avg_pool')(output)
output = Flatten(name='flatten')(output)
output = Dense(nb_classes, activation='softmax', name='predictions')(output)

InceptionV3_model = Model(InceptionV3_notop.input, output)
optimizer = SGD(lr = learning_rate, momentum = 0.9, decay = 0.0, nesterov = True)
InceptionV3_model.compile(loss='categorical_crossentropy', optimizer = optimizer, metrics = ['accuracy'])

best_model_file = "./weights.h5"
best_model = ModelCheckpoint(best_model_file, monitor='val_acc', verbose = 1, save_best_only = True)

# 数据扩增
train_datagen = ImageDataGenerator(
rescale=1./255,
shear_range=0.1,
zoom_range=0.1,
rotation_range=10.,
width_shift_range=0.1,
height_shift_range=0.1,
horizontal_flip=True)

# this is the augmentation configuration we will use for validation:
# only rescaling
val_datagen = ImageDataGenerator(rescale=1./255)

train_generator = train_datagen.flow_from_directory(
train_data_dir,
target_size = (img_width, img_height),
batch_size = batch_size,
shuffle = True,
classes = DogNames,
class_mode = 'categorical'
)
print train_generator.class_indices

validation_generator = val_datagen.flow_from_directory(
val_data_dir,
target_size=(img_width, img_height),
batch_size=batch_size,
shuffle = True,
classes = DogNames,
class_mode = 'categorical'
)

InceptionV3_model.fit_generator(
train_generator,
samples_per_epoch = nbr_train_samples,
nb_epoch = nbr_epochs,
validation_data = validation_generator,
nb_val_samples = nbr_validation_samples,
callbacks = [best_model])

完成模型训练之后，直接测试错误率在0.21，此时对测试数据通过预处理生成多个图片进行测试，提升到0.20。

# test data generator for prediction
test_datagen = ImageDataGenerator(
rescale=1./255,
shear_range=0.1,
zoom_range=0.1,
width_shift_range=0.1,
height_shift_range=0.1,
horizontal_flip=True)

print('Loading model and weights from training process ...')
InceptionV3_model = load_model(weights_path)

for idx in range(nbr_augmentation):
print('{}th augmentation for testing ...'.format(idx))
random_seed = np.random.random_integers(0, 100000)

test_generator = test_datagen.flow_from_directory(
test_data_dir,
target_size=(img_width, img_height),
batch_size=batch_size,
shuffle = False, # Important !!!
seed = random_seed,
classes = None,
class_mode = None)

test_image_list = test_generator.filenames
#print('image_list: {}'.format(test_image_list[:10]))
print('Begin to predict for testing data ...')
if idx == 0:
predictions = InceptionV3_model.predict_generator(test_generator, nbr_test_samples)
else:
predictions += InceptionV3_model.predict_generator(test_generator, nbr_test_samples)

进一步

进行度量学习，减少类内距，增大类间距，如centerloss, tripletloss。由于centerloss每次需要得到同类的特征中心，同类的狗由于不同的位姿，会导致特征中心不稳定，效果不好

将数据划分位10分，随机选取9份进行训练，一份测试，训练多个模型进行测试。

将不同分类器得到的特征进行融合，拼接在一起训练。