《国产精品操到哭》

在Altera SoC DE1板卡上跑完整的卷积神经网络

这次为(👼)大家详细展示一个利(🌉)用卷积神经网络实现图片自动(🎹)分类的例程。

神经网(💔)络的优点：自动从数据中学习经验知识，无需复杂的模型(💌)和算法。

缺点：有监督学习，需要大量的带标签数据；参数量太少时容易过拟合，泛化能力差，参数量太大时训(🎐)练收敛很慢(㊗)（有可能需要几个(📦)月到几年）。

为了克服上述缺点，人们发掘了各种计算资源，包括多核CPU、GPU、DSP、ASIC、FPGA，甚至使用模拟电路。

使用CPU实现卷积神经网络比较方便调试，但性能太差，一(🤡)般人们都选用(🏂)更快的GPU实现。目(✊)前开源的框架大多都支持GPU，如伯克利大学Caffe和Google Convnet。

微软(🧒)在2015年2月宣(🔲)布使用Stratix V完成了CNN加速器，处理 CIFAR10 图片速度可达每秒2300多张。

这里我们也使用CIFAR10图片数据，在Cyclone V板子上跑一个卷积神经网络CNN demo。由于板子上计算资源太少（DSP Slice只有(📚)80多个），实现完(😖)整的网络不太现实，只能在FPGA上实(🦉)现基本计算单元，然后由HPS统一调度。性能预期不会太高，后面给出。

CIFAR10图片都是什么呢？先来张图！

有兴趣的朋友可以到(🆖)官网下载（CIFAR10官网）。上面提到过，CNN是有监督(🥕)学习系统，需要大量带label的数据，CIFAR10就是这样一个开放的数据库，提供了60000张不同类别的图片(⌚)，分为10个类（如上图左侧所示），每个类别有600张图。这个数据集不(🍤)算特别大，适合在嵌(🐥)入式平台上实现。而更大的数据集有ImageNet-1000（ImageNet官网），拥有120多万张高清无码大图，我下载到硬盘，占用了近200GB空间（只能忍痛将其他rmvb和avi删掉了）！

有朋友会问，不用这些数据(🔶)行不行，我们的智能手机里面照片能不能用于CNN做训练？

答(🔵)案是可以的，只是你的数据集很不“均匀”，采样不(🏉)够“完备”，训练出的模型是真实模型的“有偏估计”，而上述两个数据集经过了种种考验(🚝)，已经是学术界公认的优质数据集，一年一度的ILSVRC比赛就采用了这些数据集。

说完数据(🦓)，再说(🎹)模型。先来看一张经典的CNN结构：

这是世界上第一个将CNN实用化的例子，实现了手写体字母自动识别。在这个CNN模型(✖)中，可以看到输入是一张32 x 32的二维图像，经过卷(🐴)积(🌵)层(Convolution)、下采样层（Subsampling，也称Pooling)、全连接层（Full Connection，也称Inner Product)后，得到一组概率密度，我们选其中概率最大的元(😯)素作为该模型(🍻)对输入图像的分类结果。所以实现CNN时，只需要实现三种基本算法：卷(💋)积、下采样、矩(🥓)阵乘。除此之外，每层输出都可选择是否经过非线性变换，常用的非线性变换有ReLU和(🥒)Sigmoid，前者计算较为简单，使用较为广泛。

Caffe框架中提供了专门为cifar10数据定制的模型，是用(🏤)proto格式写的，我们的demo也基于这个模型。内容如下(😝)：

name: "CIFAR10_quick_test"input: "data"input_dim: 1input_dim: 3input_dim: 32input_dim: 32layers {name: "conv1"type: CONVOLUTIONbottom: "data"top: "conv1"blobs_lr: 1blobs_lr: 2convolution_param {num_output: 32pad: 2kernel_size: 5stride: 1}}layers {name: "pool1"type: POOLINGbottom: "conv1"top: "pool1"pooling_param {pool: MAXkernel_size: 3stride: 2}}layers {name: "relu1"type: RELUbottom: "pool1"top: "pool1"}layers {name: "conv2"type: CONVOLUTIONbottom: "pool1"top: "conv2"blobs_lr: 1blobs_lr: 2convolution_param {num_output: 32pad: 2kernel_size: 5stride: 1}}layers {name: "relu2"type: RELUbottom: "conv2"top: "conv2"}layers {name: "pool2"type: POOLINGbottom: "conv2"top: "pool2"pooling_param {pool: AVEkernel_size: 3stride: 2}}layers {name: "conv3"type: CONVOLUTIONbottom: "pool2"top: "conv3"blobs_lr: 1blobs_lr: 2convolution_param {num_output: 64pad: 2kernel_size: 5stride: 1}}layers {name: "relu3"type: RELUbottom: "conv3"top: "conv3"}layers {name: "pool3"type: POOLINGbottom: "conv3"top: "pool3"pooling_param {pool: AVEkernel_size: 3stride: 2}}layers {name: "ip1"type: INNER_PRODUCTbottom: "pool3"top: "ip1"blobs_lr: 1blobs_lr: 2inner_product_param {num_output: 64}}layers {name: "ip2"type: INNER_PRODUCTbottom: "ip1"top: "ip2"blobs_lr: 1blobs_lr: 2inner_product_param {num_output: 10}}layers {name: "prob"type: SOFTMAXbottom: "ip2"top: "prob"}

复制代码

可见，上述模型经(💮)过了3个卷积(😀)层（conv1, conv2, conv3），每个卷积层后面都跟着(🚳)下采样层（pool1, pool2, pool3），之后有两个全连接层(ip1, ip2)，最后一层prob为SOFTMAX分类层，是计算概率(🎬)密度的，这里我们(💠)不需要关心。

下面三张图分别统计了CNN模型各层的参数量、数(🐅)据量和计算量。

可以看出，卷积层的参数量很少，但(🚘)数据量很大；全连接层刚好相反，参数量较大，但数(🔁)据(🍢)量很少。

通过计算量统计发现conv2计(🐘)算量最大，其次是conv3和conv1。全连接层的计算量相对卷积层(👦)较小，但不可忽略。其他层（pool1, pool2以及各级relu）由于计算量太小，本设计中(🥌)没有将(🤮)其实(📐)现为Open CL kernel，而是直接CPU端实现。

综上(🥠)所述，我们重点实现两个算法：卷积和矩阵乘，分别对应卷积层、全连接层的实现。

在DE1-SOC上我利用了友晶提供的Open CL BSP，支持C语言开发FPGA。

卷积层计算kernel函数如下：

__attribute__((num_compute_units(4)))__kernelvoid conv(__global float * a, __global float * b, __global float * c, const int M, const int N, const int K){int gx = get_global_id(0);int gy = get_global_id(1);float tmp=0.0f;for(int x = 0; x < K; x ++){for(int y = 0; y < K; y ++){tmp += a[(gx + x) * M + (gy + y)] * b[x * K + y];}}

复制代码

全连接层计算采用矩阵乘实现，kernel函数如下：

__attribute__((num_compute_units(4)))__kernelvoid gemm(__global float * a, __global float * b, __global float * c, const int M, const int N, const int K){int gx = get_global_id(0);int gy = get_global_id(1);int sy = get_global_size(1);int sx = get_global_size(0);int s = sx * sy;for(int x = gx; x < M; x += sx){for(int y = gy; y < N; y += sy){float tmp=0.0f;for(int z = 0; z < K; z++){tmp += a[z * M + x] * b[y * K + z];}c[y * M + x] = tmp;}}}

复制代码

编译kernel函数需(📍)要使用Altera SDK for OpenCL，我用的(🎽)版本是14.0.0.200，申请了两个月的license。编译使用命令行aoc，得到*.aocx文件。

Open CL编译输(🌧)出报(🦁)告中给出了资源占用情(🚟)况：

+--------------------------------------------------------------------+; Estimated Resource Usage Summary ;+----------------------------------------+---------------------------+; Resource + Usage ;+----------------------------------------+---------------------------+; Logic utilization ; 83% ;; Dedicated logic registers ; 46% ;; Memory blocks ; 57% ;; DSP blocks ; 25% ;+----------------------------------------+---------------------------;

复制代码

可见，逻辑资源、(🍤)存储器(🌘)资(🌬)源消耗较为明显，而DSP资源并未用尽，说明还有优化(📰)的空(🏬)间。

编(🐎)译主程序需要使用SoCEDS，我用的版本为(🚊)14.0.2.274，也是命令行方式，在工程目录下执行make，结束后得(🛂)到可执行文件cnn。

将这两个文件拷贝到SD卡，按照前面的博客对板子进行设置，将CNN的(🌷)模型、CIFAR10数据也拷贝到SD卡中，板子上电，mount SD卡到/mnt，执行cnn，得到输出如下：

<div class="blockcode"><blockquote>Please input the number of images(1~100):100Loading data...OK!Constructing CNN...OK!Begin calculation...Elapsed Time = 141.861 s.Real Label = 3(cat), Calc Label = 3(cat), error count = 0Real Label = 8(ship), Calc Label = 8(ship), error count = 0Real Label = 8(ship), Calc Label = 8(ship), error count = 0Real Label = 0(airplane), Calc Label = 0(airplane), error count = 0Real Label = 6(frog), Calc Label = 6(frog), error count = 0Real Label = 6(frog), Calc Label = 6(frog), error count = 0Real Label = 1(automobile), Calc Label = 1(automobile), error count = 0Real Label = 6(frog), Calc Label = 6(frog), error count = 0Real Label = 3(cat), Calc Label = 3(cat), error count = 0Real Label = 1(automobile), Calc Label = 1(automobile), error count = 0Real Label = 0(airplane), Calc Label = 0(airplane), error count = 0Real Label = 9(truck), Calc Label = 9(truck), error count = 0Real Label = 5(dog), Calc Label = 5(dog), error count = 0Real Label = 7(horse), Calc Label = 7(horse), error count = 0Real Label = 9(truck), Calc Label = 9(truck), error count = 0Real Label = 8(ship), Calc Label = 8(ship), error count = 0Real Label = 5(dog), Calc Label = 5(dog), error count = 0Real Label = 7(horse), Calc Label = 7(horse), error count = 0Real Label = 8(ship), Calc Label = 8(ship), error count = 0Real Label = 6(frog), Calc Label = 6(frog), error count = 0Real Label = 7(horse), Calc Label = 7(horse), error count = 0Real Label = 0(airplane), Calc Label = 2(bird), error count = 1Real Label = 4(deer), Calc Label = 4(deer), error count = 1Real Label = 9(truck), Calc Label = 9(truck), error count = 1Real Label = 5(dog), Calc Label = 4(deer), error count = 2Real Label = 2(bird), Calc Label = 3(cat), error count = 3Real Label = 4(deer), Calc Label = 4(deer), error count = 3Real Label = 0(airplane), Calc Label = 0(airplane), error count = 3Real Label = 9(truck), Calc Label = 9(truck), error count = 3Real Label = 6(frog), Calc Label = 6(frog), error count = 3Real Label = 6(frog), Calc Label = 6(frog), error count = 3Real Label = 5(dog), Calc Label = 5(dog), error count = 3Real Label = 4(deer), Calc Label = 4(deer), error count = 3Real Label = 5(dog), Calc Label = 5(dog), error count = 3Real Label = 9(truck), Calc Label = 9(truck), error count = 3Real Label = 2(bird), Calc Label = 3(cat), error count = 4Real Label = 4(deer), Calc Label = 7(horse), error count = 5Real Label = 1(automobile), Calc Label = 9(truck), error count = 6Real Label = 9(truck), Calc Label = 9(truck), error count = 6Real Label = 5(dog), Calc Label = 5(dog), error count = 6Real Label = 4(deer), Calc Label = 4(deer), error count = 6Real Label = 6(frog), Calc Label = 6(frog), error count = 6Real Label = 5(dog), Calc Label = 5(dog), error count = 6Real Label = 6(frog), Calc Label = 6(frog), error count = 6Real Label = 0(airplane), Calc Label = 0(airplane), error count = 6Real Label = 9(truck), Calc Label = 9(truck), error count = 6Real Label = 3(cat), Calc Label = 5(dog), error count = 7Real Label = 9(truck), Calc Label = 9(truck), error count = 7Real Label = 7(horse), Calc Label = 7(horse), error count = 7Real Label = 6(frog), Calc Label = 6(frog), error count = 7Real Label = 9(truck), Calc Label = 9(truck), error count = 7Real Label = 8(ship), Calc Label = 8(ship), error count = 7Real Label = 0(airplane), Calc Label = 2(bird), error count = 8Real Label = 3(cat), Calc Label = 3(cat), error count = 8Real Label = 8(ship), Calc Label = 8(ship), error count = 8Real Label = 8(ship), Calc Label = 8(ship), error count = 8Real Label = 7(horse), Calc Label = 7(horse), error count = 8Real Label = 7(horse), Calc Label = 7(horse), error count = 8Real Label = 4(deer), Calc Label = 3(cat), error count = 9Real Label = 6(frog), Calc Label = 3(cat), error count = 10Real Label = 7(horse), Calc Label = 7(horse), error count = 10Real Label = 3(cat), Calc Label = 5(dog), error count = 11Real Label = 6(frog), Calc Label = 6(frog), error count = 11Real Label = 3(cat), Calc Label = 3(cat), error count = 11Real Label = 6(frog), Calc Label = 6(frog), error count = 11Real Label = 2(bird), Calc Label = 2(bird), error count = 11Real Label = 1(automobile), Calc Label = 1(automobile), error count = 11Real Label = 2(bird), Calc Label = 2(bird), error count = 11Real Label = 3(cat), Calc Label = 3(cat), error count = 11Real Label = 7(horse), Calc Label = 9(truck), error count = 12Real Label = 2(bird), Calc Label = 2(bird), error count = 12Real Label = 6(frog), Calc Label = 6(frog), error count = 12Real Label = 8(ship), Calc Label = 8(ship), error count = 12Real Label = 8(ship), Calc Label = 8(ship), error count = 12Real Label = 0(airplane), Calc Label = 0(airplane), error count = 12Real Label = 2(bird), Calc Label = 2(bird), error count = 12Real Label = 9(truck), Calc Label = 0(airplane), error count = 13Real Label = 3(cat), Calc Label = 3(cat), error count = 13Real Label = 3(cat), Calc Label = 2(bird), error count = 14Real Label = 8(ship), Calc Label = 8(ship), error count = 14Real Label = 8(ship), Calc Label = 8(ship), error count = 14Real Label = 1(automobile), Calc Label = 1(automobile), error count = 14Real Label = 1(automobile), Calc Label = 1(automobile), error count = 14Real Label = 7(horse), Calc Label = 7(horse), error count = 14Real Label = 2(bird), Calc Label = 2(bird), error count = 14Real Label = 5(dog), Calc Label = 7(horse), error count = 15Real Label = 2(bird), Calc Label = 2(bird), error count = 15Real Label = 7(horse), Calc Label = 7(horse), error count = 15Real Label = 8(ship), Calc Label = 8(ship), error count = 15Real Label = 9(truck), Calc Label = 9(truck), error count = 15Real Label = 0(airplane), Calc Label = 0(airplane), error count = 15Real Label = 3(cat), Calc Label = 4(deer), error count = 16Real Label = 8(ship), Calc Label = 8(ship), error count = 16Real Label = 6(frog), Calc Label = 6(frog), error count = 16Real Label = 4(deer), Calc Label = 4(deer), error count = 16Real Label = 6(frog), Calc Label = 6(frog), error count = 16Real Label = 6(frog), Calc Label = 6(frog), error count = 16Real Label = 0(airplane), Calc Label = 2(bird), error count = 17Real Label = 0(airplane), Calc Label = 0(airplane), error count = 17Real Label = 7(horse), Calc Label = 7(horse), error count = 17Classify Score = 83 %.

上面的(🐇)执行流程是这样的，首先(🎭)输入测试样本数目（1到100），由于DE1板子FPGA端SDRAM容量较小，难以加载全部测试数据（10000张图片），故每次最多装入100张图片。之后(📰)载入数据到HPS内存，然后开始构建CNN模型，构建过程中(📥)也实现了Open CL的初始化。构建完(🍉)毕，将输入图(🚓)像依次(🐾)通过CNN，得(🔒)到一系列分类结果，与标签进行对比，统计错误(🦅)分类个数，计算分类准确率。

经过测试，分类准确率达到83%，与Caffe测试结果一致。

经过以上测试，可以得到结(🏯)论：

（1）使用Open CL可以很方便地移植高级语言编写的(➖)算法；

（2）CNN在移植过程中需要考虑实际硬件，定制合适的模型和数据；

（(🏙)3）Cyclone 5逻辑资源(🈸)较少（85K，Open CL kernel占用了(💼)83%），如果希望进一步提高计算速度，一方面可以选用高性能器件（如Stratix V、Arria 10），另一方面可以(🌐)使(🐐)用RTL自己搭建计算系统。

以上图文内容均是EEWORLD论坛网友zhaoyongke原创，在此感谢。

欢迎(🏅)微博@EEWORLD

如果你也写过(💗)此类原创干货请关注微信订阅号（ID：eeworldbbs，将你的原创发至：bbs_service@eeworld.com.cn，一(🥘)经入选，我们将帮你登上头条！

与更多行业内网友进行交流请登陆EEWORLD论坛。

【国产精品操到哭的相关新闻】