Darknet源码阅读notes

darknet notes

some functions

char *basecfg(char *cfgfile)
{
    char *c = cfgfile;
    char *next;
    while((next = strchr(c, '/')))
    {
        c = next+1;
    }
    c = copy_string(c);
    next = strchr(c, '.');
    if (next) *next = 0;
    return c;
}

C 库函数 char *strchr(const char *str, int c) 在参数 str 所指向的字符串中搜索第一次出现字符 c（一个无符号字符）的位置。返回的next实际上是指向c的指针，用%s打出来也就是c（包含）之后的字符串。while 当还有斜杠时，继续往后找，没有斜杠就退出，也就是指向最后一个斜杠。与点号匹配去后缀名。

typedef struct network{
    int n;
    int batch;
    size_t *seen;
    int *t;
    float epoch;
    int subdivisions;
    layer *layers;
    float *output;
    learning_rate_policy policy;

    float learning_rate;
    float momentum;
    float decay;
    float gamma;
    float scale;
    float power;
    int time_steps;
    int step;
    int max_batches;
    float *scales;
    int   *steps;
    int num_steps;
    int burn_in;

    int adam;
    float B1;
    float B2;
    float eps;

    int inputs;
    int outputs;
    int truths;
    int notruth;
    int h, w, c;
    int max_crop;
    int min_crop;
    float max_ratio;
    float min_ratio;
    int center;
    float angle;
    float aspect;
    float exposure;
    float saturation;
    float hue;
    int random;

    int gpu_index;
    tree *hierarchy;

    float *input;
    float *truth;
    float *delta;
    float *workspace;
    int train;
    int index;
    float *cost;
    float clip;

#ifdef GPU
    float *input_gpu;
    float *truth_gpu;
    float *delta_gpu;
    float *output_gpu;
#endif

} network;

darknet.h中对network的定义。

network *load_network(char *cfg, char *weights, int clear)
{
    network *net = parse_network_cfg(cfg);
    if(weights && weights[0] != 0){
        load_weights(net, weights);
    }
    if(clear) (*net->seen) = 0;
    return net;
}

通过cfg文件和已有的权重加载网络模型。从cfg到网络的过程有函数parse_network_cfg实现。首先看如何读入cfg文件到list* 类型。

list *read_cfg(char *filename)
{
    FILE *file = fopen(filename, "r");
    if(file == 0) file_error(filename);
    char *line;
    int nu = 0;
    list *options = make_list();
    section *current = 0;
    while((line=fgetl(file)) != 0){
        ++ nu;
        strip(line);
        switch(line[0]){
            case '[':
                current = malloc(sizeof(section));
                list_insert(options, current);
                current->options = make_list();
                current->type = line;
                break;
            case '':
            case '#':
            case ';':
                free(line);
                break;
            default:
                if(!read_option(line, current->options)){
                    fprintf(stderr, "Config file error line %d, could parse: %s
", nu, line);
                    free(line);
                }
                break;
        }
    }
    fclose(file);
    return options;
}

可以看出，先判断每一行有没有内容。这里的read_cfg读到一个[ ]中括号，就开一块内存，并且把它加入到现有的options的list中。遇到反斜杠0，#或；都直接break。如果不是按照规则写的则报错。返回options。

下面再看parse_network_cfg

network *parse_network_cfg(char *filename)
{
    list *sections = read_cfg(filename);
    node *n = sections->front;
    if(!n) error("Config file has no sections");
    network *net = make_network(sections->size - 1);
    net->gpu_index = gpu_index;
    size_params params;

    section *s = (section *)n->val;
    list *options = s->options;
    if(!is_network(s)) error("First section must be [net] or [network]");
    parse_net_options(options, net);

    params.h = net->h;
    params.w = net->w;
    params.c = net->c;
    params.inputs = net->inputs;
    params.batch = net->batch;
    params.time_steps = net->time_steps;
    params.net = net;

    size_t workspace_size = 0;
    n = n->next;
    int count = 0;
    free_section(s);
    fprintf(stderr, "layer     filters    size              input                output
");
    while(n){
        params.index = count;
        fprintf(stderr, "%5d ", count);
        s = (section *)n->val;
        options = s->options;
        layer l = {0};
        LAYER_TYPE lt = string_to_layer_type(s->type);
        if(lt == CONVOLUTIONAL){
            l = parse_convolutional(options, params);
        }else if(lt == DECONVOLUTIONAL){
            l = parse_deconvolutional(options, params);
        }else if(lt == LOCAL){
            l = parse_local(options, params);
        }else if(lt == ACTIVE){
            l = parse_activation(options, params);
        }else if(lt == LOGXENT){
            l = parse_logistic(options, params);
        }else if(lt == L2NORM){
            l = parse_l2norm(options, params);
        }else if(lt == RNN){
            l = parse_rnn(options, params);
        }else if(lt == GRU){
            l = parse_gru(options, params);
        }else if (lt == LSTM) {
            l = parse_lstm(options, params);
        }else if(lt == CRNN){
            l = parse_crnn(options, params);
        }else if(lt == CONNECTED){
            l = parse_connected(options, params);
        }else if(lt == CROP){
            l = parse_crop(options, params);
        }else if(lt == COST){
            l = parse_cost(options, params);
        }else if(lt == REGION){
            l = parse_region(options, params);
        }else if(lt == YOLO){
            l = parse_yolo(options, params);
        }else if(lt == DETECTION){
            l = parse_detection(options, params);
        }else if(lt == SOFTMAX){
            l = parse_softmax(options, params);
            net->hierarchy = l.softmax_tree;
        }else if(lt == NORMALIZATION){
            l = parse_normalization(options, params);
        }else if(lt == BATCHNORM){
            l = parse_batchnorm(options, params);
        }else if(lt == MAXPOOL){
            l = parse_maxpool(options, params);
        }else if(lt == REORG){
            l = parse_reorg(options, params);
        }else if(lt == AVGPOOL){
            l = parse_avgpool(options, params);
        }else if(lt == ROUTE){
            l = parse_route(options, params, net);
        }else if(lt == UPSAMPLE){
            l = parse_upsample(options, params, net);
        }else if(lt == SHORTCUT){
            l = parse_shortcut(options, params, net);
        }else if(lt == DROPOUT){
            l = parse_dropout(options, params);
            l.output = net->layers[count-1].output;
            l.delta = net->layers[count-1].delta;
#ifdef GPU
            l.output_gpu = net->layers[count-1].output_gpu;
            l.delta_gpu = net->layers[count-1].delta_gpu;
#endif
        }else{
            fprintf(stderr, "Type not recognized: %s
", s->type);
        }
        l.clip = net->clip;
        l.truth = option_find_int_quiet(options, "truth", 0);
        l.onlyforward = option_find_int_quiet(options, "onlyforward", 0);
        l.stopbackward = option_find_int_quiet(options, "stopbackward", 0);
        l.dontsave = option_find_int_quiet(options, "dontsave", 0);
        l.dontload = option_find_int_quiet(options, "dontload", 0);
        l.dontloadscales = option_find_int_quiet(options, "dontloadscales", 0);
        l.learning_rate_scale = option_find_float_quiet(options, "learning_rate", 1);
        l.smooth = option_find_float_quiet(options, "smooth", 0);
        option_unused(options);
        net->layers[count] = l;
        if (l.workspace_size > workspace_size) workspace_size = l.workspace_size;
        free_section(s);
        n = n->next;
        ++count;
        if(n){
            params.h = l.out_h;
            params.w = l.out_w;
            params.c = l.out_c;
            params.inputs = l.outputs;
        }
    }
    free_list(sections);
    layer out = get_network_output_layer(net);
    net->outputs = out.outputs;
    net->truths = out.outputs;
    if(net->layers[net->n-1].truths) net->truths = net->layers[net->n-1].truths;
    net->output = out.output;
    net->input = calloc(net->inputs*net->batch, sizeof(float));
    net->truth = calloc(net->truths*net->batch, sizeof(float));
#ifdef GPU
    net->output_gpu = out.output_gpu;
    net->input_gpu = cuda_make_array(net->input, net->inputs*net->batch);
    net->truth_gpu = cuda_make_array(net->truth, net->truths*net->batch);
#endif
    if(workspace_size){
        //printf("%ld
", workspace_size);
#ifdef GPU
        if(gpu_index >= 0){
            net->workspace = cuda_make_array(0, (workspace_size-1)/sizeof(float)+1);
        }else {
            net->workspace = calloc(1, workspace_size);
        }
#else
        net->workspace = calloc(1, workspace_size);
#endif
    }
    return net;
}

主要内容就是通过读出来的每个LAYERTYPE调用不同的parse某种layer 的函数，生成网络结构，并且把参数如batch等传进net。举栗子：parse_convolutional如下：

convolutional_layer parse_convolutional(list *options, size_params params)
{
    int n = option_find_int(options, "filters",1);
    int size = option_find_int(options, "size",1);
    int stride = option_find_int(options, "stride",1);
    int pad = option_find_int_quiet(options, "pad",0);
    int padding = option_find_int_quiet(options, "padding",0);
    int groups = option_find_int_quiet(options, "groups", 1);
    if(pad) padding = size/2;

    char *activation_s = option_find_str(options, "activation", "logistic");
    ACTIVATION activation = get_activation(activation_s);

    int batch,h,w,c;
    h = params.h;
    w = params.w;
    c = params.c;
    batch=params.batch;
    if(!(h && w && c)) error("Layer before convolutional layer must output image.");
    int batch_normalize = option_find_int_quiet(options, "batch_normalize", 0);
    int binary = option_find_int_quiet(options, "binary", 0);
    int xnor = option_find_int_quiet(options, "xnor", 0);

    convolutional_layer layer = make_convolutional_layer(batch,h,w,c,n,groups,size,stride,padding,activation, batch_normalize, binary, xnor, params.net->adam);
    layer.flipped = option_find_int_quiet(options, "flipped", 0);
    layer.dot = option_find_float_quiet(options, "dot", 0);

    return layer;
}

（to be continued）

2018年7月11日17:49:57