经典网络LeNet-5介绍及代码测试(Caffe, MNIST, C++)

清疚 2022-03-10 09:14 619阅读 0赞

LeNet-5：包含7个层(layer)，如下图所示：输入层没有计算在内，输入图像大小为32\*32\*1，是针对灰度图进行训练和预测的。论文名字为” Gradient-Based Learning Applied to Document Recognition”，可以直接从[http://yann.lecun.com/exdb/publis/pdf/lecun-01a.pdf][http_yann.lecun.com_exdb_publis_pdf_lecun-01a.pdf] 下载原始论文。

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70][]

第一层是卷积层，使用6个5\*5的filter，stride为1，padding为0，输出结果为28\*28\*6，6个feature maps，训练参数(5\*5\*1)\*6+6=156(weights + bias)；

第二层进行平均池化操作，filter为2\*2，stride为2，padding为0，输出结果为14\*14\*6，6个feature maps，训练参数1\*6+6=12(coefficient + bias)；

第三层是卷积层，使用16个5\*5的filter，stride为1，padding为0，输出结果为10\*10\*16，16个feature maps， 训练参数(按照论文给的连接方式) (5\*5\*3+1)\*6 + (5\*5\*4+1)\*6+(5\*5\*4+1)\*3+(5\*5\*6+1)\*1 = 1516(weights + bias)；

第四层又是平均池化层，filter为2\*2，stride为2，padding为0，输出结果为5\*5\*16，16个feature maps，训练参数1\*16+16=32(coefficient + bias)；

第五层是卷积层，使用120个5\*5的fiter,stride为1，输出结果为1\*1\*120，120个feature maps，训练参数(5\*5\*6)\*120+120=48120(weights + bias)；

第六层是一个全连接层，有84个神经元，训练参数120\*84+84=10164(weights + bias)；此layer使用的激活函数为tanh。

第七层得到最后的输出预测y’的值，y’有10个可能的值，对应识别0--9这10个数字，在现在的版本中，使用softmax函数输出10种分类结果，即第七层为softmax。

输入图像size为32\*32比训练数据集图像size为28\*28大的原因：期望诸如笔划终点(stroke end-points)或角点(corner)这些潜在的特征(potential distinctive feature)能够出现在最高层特征检测器(highest-level feature detectors)的感受野的中心。

没有把layer2中的每个feature map连接到layer3中的每个feature map原因：(1). 不完全的连接机制将连接的数量保持在合理的范围内；(2). 更重要的，它强制破坏了网络的对称性。因为不同的feature maps来自不同的输入，所以不同的feature maps被强制提取不同的features.(The main reason is to break the symmetry in the network and keeps the number of connections within reasonable bounds.)

LeNet-5中的5是指5个隐藏层即卷积层、池化层、卷积层、池化层、卷积层。

不同的filter可以提取不同的特征，如边沿、线性、角等特征。

关于卷积神经网络的基础介绍可参考之前的blog：

[https://blog.csdn.net/fengbingchun/article/details/50529500][https_blog.csdn.net_fengbingchun_article_details_50529500]

[https://blog.csdn.net/fengbingchun/article/details/80262495][https_blog.csdn.net_fengbingchun_article_details_80262495]

[https://blog.csdn.net/fengbingchun/article/details/68065338][https_blog.csdn.net_fengbingchun_article_details_68065338]

[https://blog.csdn.net/fengbingchun/article/details/69001433][https_blog.csdn.net_fengbingchun_article_details_69001433]

以下是参考Caffe中的测试代码对LeNet-5网络进行测试的代码，与论文中的不同处包括：

(1). 论文中要求输入层图像大小为32\*32，这里为28\*28；

(2). 论文中第一层卷积层输出是6个feature maps，这里是20个feature maps；

(3). 论文中池化层取均值，而这里取最大值；

(4). 论文中第三层卷积层输出是16个feature maps，这里是50个feature maps，而且这里第二层的feature map是连接到第三层的每个feature map的；

(5). 论文中第五层是卷积层，这里是全连接层并输出500个神经元，激活函数采用ReLU；

(6). 论文中第七层是RBF(Euclidean Radial Basic Function)，这里采用Softmax。

以下是测试代码(lenet-5.cpp)：

#include "funset.hpp"
    #include "common.hpp"
    
    int lenet_5_mnist_train()
    {	
    #ifdef CPU_ONLY
    	caffe::Caffe::set_mode(caffe::Caffe::CPU);
    #else
    	caffe::Caffe::set_mode(caffe::Caffe::GPU);
    #endif
    
    #ifdef _MSC_VER
    	const std::string filename{ "E:/GitCode/Caffe_Test/test_data/Net/lenet-5_mnist_windows_solver.prototxt" };
    #else
    	const std::string filename{ "test_data/Net/lenet-5_mnist_linux_solver.prototxt" };
    #endif
    	caffe::SolverParameter solver_param;
    	if (!caffe::ReadProtoFromTextFile(filename.c_str(), &solver_param)) {
    		fprintf(stderr, "parse solver.prototxt fail\n");
    		return -1;
    	}
    
    	mnist_convert(); // convert MNIST to LMDB
    
    	caffe::SGDSolver<float> solver(solver_param);
    	solver.Solve();
    
    	fprintf(stdout, "train finish\n");
    
    	return 0;
    }
    
    int lenet_5_mnist_test()
    {
    #ifdef CPU_ONLY
    	caffe::Caffe::set_mode(caffe::Caffe::CPU);
    #else
    	caffe::Caffe::set_mode(caffe::Caffe::GPU);
    #endif
    
    #ifdef _MSC_VER
    	const std::string param_file{ "E:/GitCode/Caffe_Test/test_data/Net/lenet-5_mnist_windows_test.prototxt" };
    	const std::string trained_filename{ "E:/GitCode/Caffe_Test/test_data/Net/lenet-5_mnist_iter_10000.caffemodel" };
    	const std::string image_path{ "E:/GitCode/Caffe_Test/test_data/images/handwritten_digits/" };
    #else
    	const std::string param_file{ "test_data/Net/lenet-5_mnist_linux_test.prototxt" };
    	const std::string trained_filename{ "test_data/Net/lenet-5_mnist_iter_10000.caffemodel" };
    	const std::string image_path{ "test_data/images/handwritten_digits/" };
    #endif
    
    	caffe::Net<float> caffe_net(param_file, caffe::TEST);
    	caffe_net.CopyTrainedLayersFrom(trained_filename);
    
    	const boost::shared_ptr<caffe::Blob<float> > blob_data_layer = caffe_net.blob_by_name("data");
    	int image_channel_data_layer = blob_data_layer->channels();
    	int image_height_data_layer = blob_data_layer->height();
    	int image_width_data_layer = blob_data_layer->width();
    
    	const std::vector<caffe::Blob<float>*> output_blobs = caffe_net.output_blobs();
    	int require_blob_index{ -1 };
    	const int digit_category_num{ 10 };
    	for (int i = 0; i < output_blobs.size(); ++i) {
    		if (output_blobs[i]->count() == digit_category_num)
    			require_blob_index = i;
    	}
    	if (require_blob_index == -1) {
    		fprintf(stderr, "ouput blob don't match\n");
    		return -1;
    	}
    
    	std::vector<int> target{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 };
    	std::vector<int> result;
    
    	for (auto num : target) {
    		std::string str = std::to_string(num);
    		str += ".png";
    		str = image_path + str;
    
    		cv::Mat mat = cv::imread(str.c_str(), 1);
    		if (!mat.data) {
    			fprintf(stderr, "load image error: %s\n", str.c_str());
    			return -1;
    		}
    
    		if (image_channel_data_layer == 1)
    			cv::cvtColor(mat, mat, CV_BGR2GRAY);
    		else if (image_channel_data_layer == 4)
    			cv::cvtColor(mat, mat, CV_BGR2BGRA);
    
    		cv::resize(mat, mat, cv::Size(image_width_data_layer, image_height_data_layer));
    		cv::bitwise_not(mat, mat);
    
    		boost::shared_ptr<caffe::MemoryDataLayer<float> > memory_data_layer =
    			boost::static_pointer_cast<caffe::MemoryDataLayer<float>>(caffe_net.layer_by_name("data"));
    		mat.convertTo(mat, CV_32FC1, 0.00390625);
    		float dummy_label[1] {0};
    		memory_data_layer->Reset((float*)(mat.data), dummy_label, 1);
    
    		float loss{ 0.0 };
    		const std::vector<caffe::Blob<float>*>& results = caffe_net.ForwardPrefilled(&loss);
    		const float* output = results[require_blob_index]->cpu_data();
    
    		float tmp{ -1 };
    		int pos{ -1 };
    
    		for (int j = 0; j < 10; j++) {
    			//fprintf(stdout, "Probability to be Number %d is: %.3f\n", j, output[j]);
    			if (tmp < output[j]) {
    				pos = j;
    				tmp = output[j];
    			}
    		}
    
    		result.push_back(pos);
    	}
    
    	for (auto i = 0; i < 10; i++)
    		fprintf(stdout, "actual digit is: %d, result digit is: %d\n", target[i], result[i]);
    
    	fprintf(stdout, "predict finish\n");
    	return 0;
    }

solver.prototxt文件内容如下：

# solver.prototxt是一个配置文件用来告知Caffe怎样对网络进行训练
    # 其文件内的各字段名需要在caffe.proto的message SolverParameter中存在，否则会解析不成功
    
    net: "test_data/Net/lenet-5_mnist_linux_train.prototxt" # 训练网络文件名
    test_iter: 100 # test_iter * test_batch_size = 测试图像总数量
    test_interval: 500 # 指定执行多少次训练网络执行一次测试网络
    base_lr: 0.01 # 学习率
    lr_policy: "inv" # 学习策略, return base_lr * (1 + gamma * iter) ^ (- power)
    momentum: 0.9 # 动量
    weight_decay: 0.0005 # 权值衰减
    gamma: 0.0001 # 学习率计算参数
    power: 0.75 # 学习率计算参数
    display: 100 # 指定训练多少次在屏幕上显示一次结果信息，如loss值等
    max_iter: 10000 # 最多训练次数
    snapshot: 5000 # 执行多少次训练保存一次中间结果
    snapshot_prefix: "test_data/Net/lenet-5_mnist" # 结果保存位置前缀
    solver_type: SGD # 随机梯度下降

train时的prototxt文件内容如下：

name: "LeNet-5"
    
    layer {
      name: "mnist"
      type: "Data"
      top: "data"
      top: "label"
      include {
        phase: TRAIN
      }
      transform_param {
        scale: 0.00390625
      }
      data_param {
        source: "test_data/MNIST/train"
        batch_size: 64
        backend: LMDB
      }
    }
    layer {
      name: "mnist"
      type: "Data"
      top: "data"
      top: "label"
      include {
        phase: TEST
      }
      transform_param {
        scale: 0.00390625
      }
      data_param {
        source: "test_data/MNIST/test"
        batch_size: 100
        backend: LMDB
      }
    }
    layer {
      name: "conv1"
      type: "Convolution"
      bottom: "data"
      top: "conv1"
      param {
        lr_mult: 1
      }
      param {
        lr_mult: 2
      }
      convolution_param {
        num_output: 20
        kernel_size: 5
        stride: 1
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
        }
      }
    }
    layer {
      name: "pool1"
      type: "Pooling"
      bottom: "conv1"
      top: "pool1"
      pooling_param {
        pool: MAX
        kernel_size: 2
        stride: 2
      }
    }
    layer {
      name: "conv2"
      type: "Convolution"
      bottom: "pool1"
      top: "conv2"
      param {
        lr_mult: 1
      }
      param {
        lr_mult: 2
      }
      convolution_param {
        num_output: 50
        kernel_size: 5
        stride: 1
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
        }
      }
    }
    layer {
      name: "pool2"
      type: "Pooling"
      bottom: "conv2"
      top: "pool2"
      pooling_param {
        pool: MAX
        kernel_size: 2
        stride: 2
      }
    }
    layer {
      name: "ip1"
      type: "InnerProduct"
      bottom: "pool2"
      top: "ip1"
      param {
        lr_mult: 1
      }
      param {
        lr_mult: 2
      }
      inner_product_param {
        num_output: 500
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
        }
      }
    }
    layer {
      name: "relu1"
      type: "ReLU"
      bottom: "ip1"
      top: "ip1"
    }
    layer {
      name: "ip2"
      type: "InnerProduct"
      bottom: "ip1"
      top: "ip2"
      param {
        lr_mult: 1
      }
      param {
        lr_mult: 2
      }
      inner_product_param {
        num_output: 10
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
        }
      }
    }
    layer {
      name: "accuracy"
      type: "Accuracy"
      bottom: "ip2"
      bottom: "label"
      top: "accuracy"
      include {
        phase: TEST
      }
    }
    layer {
      name: "loss"
      type: "SoftmaxWithLoss"
      bottom: "ip2"
      bottom: "label"
      top: "loss"
    }

train.prototxt可视化结果如下：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70 1][]

test时的test.prototxt文件内容如下：

name: "LeNet-5"
    
    layer {
      name: "data"
      type: "MemoryData"
      top: "data" #
      top: "label"
      memory_data_param {
        batch_size: 1
        channels: 1
        height: 28
        width: 28
      }
      transform_param {
        scale: 0.00390625
      }
    }
    layer {
      name: "conv1"
      type: "Convolution"
      bottom: "data"
      top: "conv1"
      param {
        lr_mult: 1
      }
      param {
        lr_mult: 2
      }
      convolution_param {
        num_output: 20
        kernel_size: 5
        stride: 1
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
        }
      }
    }
    layer {
      name: "pool1"
      type: "Pooling"
      bottom: "conv1"
      top: "pool1"
      pooling_param {
        pool: MAX
        kernel_size: 2
        stride: 2
      }
    }
    layer {
      name: "conv2"
      type: "Convolution"
      bottom: "pool1"
      top: "conv2"
      param {
        lr_mult: 1
      }
      param {
        lr_mult: 2
      }
      convolution_param {
        num_output: 50
        kernel_size: 5
        stride: 1
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
        }
      }
    }
    layer {
      name: "pool2"
      type: "Pooling"
      bottom: "conv2"
      top: "pool2"
      pooling_param {
        pool: MAX
        kernel_size: 2
        stride: 2
      }
    }
    layer {
      name: "ip1"
      type: "InnerProduct"
      bottom: "pool2"
      top: "ip1"
      param {
        lr_mult: 1
      }
      param {
        lr_mult: 2
      }
      inner_product_param {
        num_output: 500
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
        }
      }
    }
    layer {
      name: "relu1"
      type: "ReLU"
      bottom: "ip1"
      top: "ip1"
    }
    layer {
      name: "ip2"
      type: "InnerProduct"
      bottom: "ip1"
      top: "ip2"
      param {
        lr_mult: 1
      }
      param {
        lr_mult: 2
      }
      inner_product_param {
        num_output: 10
        weight_filler {
          type: "xavier"
        }
        bias_filler {
          type: "constant"
        }
      }
    }
    layer {
      name: "prob"
      type: "Softmax"
      bottom: "ip2"
      top: "prob"
    }

test.prototxt可视化结果如下：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70 2][]

train时测试代码执行结果如下：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70 3][]

test是测试代码执行结果如下：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70 4][]

**GitHub**：[https://github.com/fengbingchun/Caffe\_Test][https_github.com_fengbingchun_Caffe_Test]

[http_yann.lecun.com_exdb_publis_pdf_lecun-01a.pdf]: http://yann.lecun.com/exdb/publis/pdf/lecun-01a.pdf
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70]: /images/20220310/f7c0743a939e406e96b2d84f41f4f63d.png
[https_blog.csdn.net_fengbingchun_article_details_50529500]: https://blog.csdn.net/fengbingchun/article/details/50529500
[https_blog.csdn.net_fengbingchun_article_details_80262495]: https://blog.csdn.net/fengbingchun/article/details/80262495
[https_blog.csdn.net_fengbingchun_article_details_68065338]: https://blog.csdn.net/fengbingchun/article/details/68065338
[https_blog.csdn.net_fengbingchun_article_details_69001433]: https://blog.csdn.net/fengbingchun/article/details/69001433
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70 1]: /images/20220310/28eb0e06dc494168967f88a551bcf7bf.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70 2]: /images/20220310/fbdf31e408c342aa995886bf02d88ff4.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70 3]: /images/20220310/c8ca349d4ac04c46b4f795578bd6a4f6.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZlbmdiaW5nY2h1bg_size_16_color_FFFFFF_t_70 4]: /images/20220310/3b3bfc35f00b458c908a3dd62b00aab1.png
[https_github.com_fengbingchun_Caffe_Test]: https://github.com/fengbingchun/Caffe_Test