【撸码caffe 二】 blob.hpp

时间 2019-11-09
标签 blob.hpp blob hpp 繁體版
原文原文链接
Blob类是caffe中对处理和传递的实际数据的封装，是caffe中基本的数据存储单元，包括前向传播中的图像数据，反向传播中的梯度数据以及网络层间的中间数据变量（包括权值，偏置等），训练模型的参数等等，能够说在caffe中，无数据不blob。api
blob能够认为是按C风格连续存储的N维数组，在硬件上能够认为是在内存中的一块连续的内存块。数组
补充一点智能指针的知识：网络
C++中的动态内存管理是经过new和delete运算符完成的，没有及时delete释放内存或者提早释放内存均可能形成内存异常，致使内存泄漏，或者是引用了非法的内存指针。app
C++ 11 标准库提供了智能指针（smart pointer）来管理动态内存对象，这种智能指针的智能之处在于能够自动释放内存对象。智能指针分为两种，一种是shared_ptr，容许多了个指针同时指向同一个对象，另外一种是unique_ptr，同时只能有一个指针指向内存对象。智能指针是模板类而不是指针，建立一个智能指针时，必须指出智能指针的对象能够指向的类型。函数
blob.hpp文件注释：ui
#ifndef CAFFE_BLOB_HPP_
#define CAFFE_BLOB_HPP_

#include <algorithm>
#include <string>
#include <vector>

#include "caffe/common.hpp"
#include "caffe/proto/caffe.pb.h"
#include "caffe/syncedmem.hpp"

//const类型的整形kMaxBlobAxes定义了Blob最大的维度，Blob的通常维度是4，图像数量*图像通道*图像宽度*图像高度
const int kMaxBlobAxes = 32;

namespace caffe {   //Blob类也定义在caffe命名空间下

	/**
	 * @brief A wrapper around SyncedMemory holders serving as the basic
	 *        computational unit through which Layer%s, Net%s, and Solver%s
	 *        interact.
	 *
	 * TODO(dox): more thorough description.
	 */
	template <typename Dtype>    //类模板
	class Blob {
	public:
		Blob()       //无参构造函数
			: data_(), diff_(), count_(0), capacity_(0) {}

		//explicit关键字的做用是防止单参数构造函数的隐式转换， 对于含有多个未初始化值的构造函数无效
		//shape是一个int型的向量，包含一个blob的维度，图像深度，高，宽信息
		//这两个构造函数都会在内部调用Reshape函数，用来设置或者修改当前blob的shape_，count_和capacity_属性
		/// @brief Deprecated; use <code>Blob(const vector<int>& shape)</code>.
		explicit Blob(const int num, const int channels, const int height,
			const int width);
		explicit Blob(const vector<int>& shape);

		/// @brief Deprecated; use <code>Reshape(const vector<int>& shape)</code>.

		//Reshape函数的做用是建立或调整blob的shape_，count_和capacity_属性
		void Reshape(const int num, const int channels, const int height,
			const int width);
		/*
		 * @brief Change the dimensions of the blob, allocating new memory if
		 *        necessary.
		 *Reshape函数能够用来建立一个初始化的内存分配信息，也能够调整前向传播过程当中网络层的数据输出尺度
		 * This function can be called both to create an initial allocation
		 * of memory, and to adjust the dimensions of a top blob during Layer::Reshape
		 blob的大小改变后，只有在已分配的内存不够的状况下才会从新分配内存， 而且新增的内存将不会被释放
		 * or Layer::Forward.When changing the size of blob, memory will only be
		 * reallocated if sufficient memory does not already exist, and excess memory
		 * will never be freed.
		 *须要注意的是直接改变输入blob的大小是错误的，它们应该在数据从低层向高层的传播中数据量增长的时候被调用
		 * Note that reshaping an input blob and immediately calling Net::Backward is
		 * an error; either Net::Forward or Net::Reshape need to be called to
		 * propagate the new input shape to higher layers.
		 */
		//经过类型为vector<int>类型的shape设置shape_，count_和capacity_ 的大小
		void Reshape(const vector<int>& shape);
		//经过类型为BlobShape类对象shape设置shape_，count_和capacity_ 的大小
		//BlobShape是在caffe.pb.h中定义的类，含有维度信息，继承自protobuf::Message
		void Reshape(const BlobShape& shape);
		//经过其余的blob对象类设置bolb的shape_，count_和capacity_ 的大小
		void ReshapeLike(const Blob& other);
		//inline定义的是内联函数，做用是将代码直接复制到调用处，节省函数调用开销，代价是增长了代码量
		//shape_string 函数用于获取bolb的打印信息(shape_和count_值)
		inline string shape_string() const {
			ostringstream stream;
			for (int i = 0; i < shape_.size(); ++i) {
				stream << shape_[i] << " ";
			}
			stream << "(" << count_ << ")";
			return stream.str();
		}
		//获取当前blob的shape_信息
		inline const vector<int>& shape() const { return shape_; }
		/**
		* @brief Returns the dimension of the index-th axis (or the negative index-th
		*        axis from the end, if index is negative).
		*
		* @param index the axis index, which may be negative as it will be
		*        "canonicalized" using CanonicalAxisIndex.
		*        Dies on out of range index.
		*/
		//获取当前指定blob指定索引的维度值
		inline int shape(int index) const {
			return shape_[CanonicalAxisIndex(index)];
		}
		//获取blob的维数
		inline int num_axes() const { return shape_.size(); }
		//获取blob的元素个数
		inline int count() const { return count_; }

		/**
		 * @brief Compute the volume of a slice; i.e., the product of dimensions
		 *        among a range of axes.
		 *
		 * @param start_axis The first axis to include in the slice.
		 *
		 * @param end_axis The first axis to exclude from the slice.
		 */
		//根据指定的开始维度和结束维度计算blob元素的个数
		inline int count(int start_axis, int end_axis) const {
			CHECK_LE(start_axis, end_axis);
			CHECK_GE(start_axis, 0);
			CHECK_GE(end_axis, 0);
			CHECK_LE(start_axis, num_axes());
			CHECK_LE(end_axis, num_axes());
			int count = 1;
			for (int i = start_axis; i < end_axis; ++i) {
				count *= shape(i);
			}
			return count;
		}
		/**
		 * @brief Compute the volume of a slice spanning from a particular first
		 *        axis to the final axis.
		 *
		 * @param start_axis The first axis to include in the slice.
		 */

		//根据指定的开始维度计算剩下的blob元素个数，内部是经过调用上边定义的count函数实现的
		inline int count(int start_axis) const {
			return count(start_axis, num_axes());
		}

		/**
		 * @brief Returns the 'canonical' version of a (usually) user-specified axis,
		 *        allowing for negative indexing (e.g., -1 for the last axis).
		 *
		 * @param axis_index the axis index.
		 *        If 0 <= index < num_axes(), return index.
		 *        If -num_axes <= index <= -1, return (num_axes() - (-index)),
		 *        e.g., the last axis index (num_axes() - 1) if index == -1,
		 *        the second to last if index == -2, etc.
		 *        Dies on out of range index.
		 */
		//CanonicalAxisIndex是用于对blob的axis_index进行转化，容许axis_index的值是负值，经过
		//CanonicalAxisIndex函数内定义的规则，转换成正值
		inline int CanonicalAxisIndex(int axis_index) const {
			CHECK_GE(axis_index, -num_axes())
				<< "axis " << axis_index << " out of range for " << num_axes()
				<< "-D Blob with shape " << shape_string();
			CHECK_LT(axis_index, num_axes())
				<< "axis " << axis_index << " out of range for " << num_axes()
				<< "-D Blob with shape " << shape_string();
			if (axis_index < 0) {
				return axis_index + num_axes();
			}
			return axis_index;
		}

		//获取当前blob的维度，推荐直接使用shape(0)获取
		/// @brief Deprecated legacy shape accessor num: use shape(0) instead.
		inline int num() const { return LegacyShape(0); }
		//获取当前blob的通道数（深度），推荐直接使用shape(1)获取
		/// @brief Deprecated legacy shape accessor channels: use shape(1) instead.
		inline int channels() const { return LegacyShape(1); }
		//获取当前blob的图像高度信息，推荐直接使用shape(2)获取
		/// @brief Deprecated legacy shape accessor height: use shape(2) instead.
		inline int height() const { return LegacyShape(2); }
		//获取当前blob的图像宽度信息，推荐直接使用shape(3)获取
		/// @brief Deprecated legacy shape accessor width: use shape(3) instead.
		inline int width() const { return LegacyShape(3); }

		//获取某一维度index下的信息，index要求在[0,3]或[-4，-1]区间内，貌似是对shape函数的封装，只是增长了形参的断定
		inline int LegacyShape(int index) const {
			CHECK_LE(num_axes(), 4)
				<< "Cannot use legacy accessors on Blobs with > 4 axes.";
			CHECK_LT(index, 4);
			CHECK_GE(index, -4);
			if (index >= num_axes() || index < -num_axes()) {
				// Axis is out of range, but still in [0, 3] (or [-4, -1] for reverse
				// indexing) -- this special case simulates the one-padding used to fill
				// extraneous axes of legacy blobs.
				return 1;
			}
			return shape(index);
		}

		//经过blob的维度n，通道（图像深度）c，图像高度h和图像宽度w计算偏移量，由该偏移量能够
		//惟必定位到bolb内的一张图像上的一个像素上
		inline int offset(const int n, const int c = 0, const int h = 0,
			const int w = 0) const {
			CHECK_GE(n, 0);
			CHECK_LE(n, num());
			CHECK_GE(channels(), 0);
			CHECK_LE(c, channels());
			CHECK_GE(height(), 0);
			CHECK_LE(h, height());
			CHECK_GE(width(), 0);
			CHECK_LE(w, width());
			return ((n * channels() + c) * height() + h) * width() + w;
		}

		//根据vector<int>类型的indices计算偏移量
		inline int offset(const vector<int>& indices) const {
			CHECK_LE(indices.size(), num_axes());
			int offset = 0;
			for (int i = 0; i < num_axes(); ++i) {
				offset *= shape(i);
				if (indices.size() > i) {
					CHECK_GE(indices[i], 0);
					CHECK_LT(indices[i], shape(i));
					offset += indices[i];
				}
			}
			return offset;
		}
		/**
		 * @brief Copy from a source Blob.
		 *
		 * @param source the Blob to copy from
		 * @param copy_diff if false, copy the data; if true, copy the diff
		 * @param reshape if false, require this Blob to be pre-shaped to the shape
		 *        of other (and die otherwise); if true, Reshape this Blob to other's
		 *        shape if necessary
		 */

		//复制外部的blob数据到本blob，根据状况拷贝data_数据仍是diff_数据，以及是否从新分配大小
		void CopyFrom(const Blob<Dtype>& source, bool copy_diff = false,
			bool reshape = false);

		//根据给定的维度，深度，宽高位置信息获取前向传播数据的一个元素的值
		inline Dtype data_at(const int n, const int c, const int h,
			const int w) const {
			return cpu_data()[offset(n, c, h, w)];
		}

		//根据给定的维度，深度，宽高位置信息获取反向传播梯度diff_的一个元素的值
		inline Dtype diff_at(const int n, const int c, const int h,
			const int w) const {
			return cpu_diff()[offset(n, c, h, w)];
		}

		//获取前向传播数据data_的指针
		inline Dtype data_at(const vector<int>& index) const {
			return cpu_data()[offset(index)];
		}

		//获取反向传播梯度diff_的指针
		inline Dtype diff_at(const vector<int>& index) const {
			return cpu_diff()[offset(index)];
		}

		//获取前向传播数据对象
		inline const shared_ptr<SyncedMemory>& data() const {
			CHECK(data_);
			return data_;
		}

		//获取反向传播梯度对象
		inline const shared_ptr<SyncedMemory>& diff() const {
			CHECK(diff_);
			return diff_;
		}

		const Dtype* cpu_data() const;     //定义的获取cpu数据指针函数
		void set_cpu_data(Dtype* data);   //设置cpu数据
		const int* gpu_shape() const;       //返回GPU shape_数据指针
		const Dtype* gpu_data() const;    //返回GPU 数据指针
		const Dtype* cpu_diff() const;      //返回CPU上反向传播的梯度数据指针
		const Dtype* gpu_diff() const;      //返回GPU上反向传播的梯度数据指针
		Dtype* mutable_cpu_data();          //如下加上mutable表明能够修改获取到的数据
		Dtype* mutable_gpu_data();
		Dtype* mutable_cpu_diff();
		Dtype* mutable_gpu_diff();

		//梯度降低过程当中训练参数更新
		void Update();
		//从BlobProto中导入数据到当前blob，完成数据解析（反序列化）
		void FromProto(const BlobProto& proto, bool reshape = true);
		//把blob数据导入BlobProto，完成数据序列化
		void ToProto(BlobProto* proto, bool write_diff = false) const;

		//计算data_的L1范式：向量中各个元素绝对值之和
		/// @brief Compute the sum of absolute values (L1 norm) of the data.
		Dtype asum_data() const;

		//计算diff_的L1范式：向量中各个元素绝对值之和
		/// @brief Compute the sum of absolute values (L1 norm) of the diff.
		Dtype asum_diff() const;

		//计算data_的L2范式：向量中各个元素的平方和
		/// @brief Compute the sum of squares (L2 norm squared) of the data.
		Dtype sumsq_data() const;

		//计算diff_的L2范式：向量中各个元素的平方和
		/// @brief Compute the sum of squares (L2 norm squared) of the diff.
		Dtype sumsq_diff() const;

		//将data_数据乘以一个系数scale_factor
		/// @brief Scale the blob data by a constant factor.
		void scale_data(Dtype scale_factor);

		//将diff _数据乘以一个系数scale_factor
		/// @brief Scale the blob diff by a constant factor.
		void scale_diff(Dtype scale_factor);

		/**
		 * @brief Set the data_ shared_ptr to point to the SyncedMemory holding the
		 *        data_ of Blob other -- useful in Layer%s which simply perform a copy
		 *        in their Forward pass.
		 *
		 * This deallocates the SyncedMemory holding this Blob's data_, as
		 * shared_ptr calls its destructor when reset with the "=" operator.
		 */

		//将外部一个Blob对象的数据指针指向当前的blob的数据data_，从而实现数据共享
		void ShareData(const Blob& other);
		/**
		 * @brief Set the diff_ shared_ptr to point to the SyncedMemory holding the
		 *        diff_ of Blob other -- useful in Layer%s which simply perform a copy
		 *        in their Forward pass.
		 *
		 * This deallocates the SyncedMemory holding this Blob's diff_, as
		 * shared_ptr calls its destructor when reset with the "=" operator.
		 */

		//将外部一个Blob对象的梯度指针指向当前的blob的数据diff_，从而实现数据共享
		void ShareDiff(const Blob& other);

		//判断当前blob的shape_和BlobProto中的shape_是否相同
		bool ShapeEquals(const BlobProto& other);

	protected:
		//后缀加上_表示是Blob的成员变量
		shared_ptr<SyncedMemory> data_;      //前向传播数据
		shared_ptr<SyncedMemory> diff_;      //反向传播梯度（误差）数据
		shared_ptr<SyncedMemory> shape_data_;   //blob的训练数据
		vector<int> shape_;    //blob的训练数据的组织维度
		int count_;      //blob中全部元素的个数，值为shape_中4个参数的乘积
		int capacity_;  //blob的容积量

		//禁用Blob类的拷贝和赋值操做
		DISABLE_COPY_AND_ASSIGN(Blob);
	};  // class Blob

}  // namespace caffe

#endif  // CAFFE_BLOB_HPP_