darknet-解析网络配置cfg文件

darknet中网络配置文件是cfg文件,类似caffe的train_val.prototxt文件,代码中解析这个cfg文件的函数是parse_network_cfg_custom().

这个函数列举了darknet支持的所有类型的层,以后会慢慢研究,这里重点关注params变量,它把所有层组织成一个网络.

typedef struct size_params{
    int batch; 
    int inputs; //batch * c * w * h, 当前层的输入,是前一层的输出
    int h;
    int w;
    int c;
    int index; //当前层索引
    int time_steps;
    network net; //整个网络
} size_params;
size_params params;

params变量主要用来存储前一层的尺寸信息,帮助当前层的建立.

network parse_network_cfg_custom(char *filename, int batch)
{
    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;
    if (batch > 0) net.batch = batch;
    params.batch = net.batch;
    params.time_steps = net.time_steps;
    params.net = net;

    float bflops = 0;
    size_t workspace_size = 0;
    n = n->next;
    int count = 0;
    free_section(s);
    fprintf(stderr, "layer     filters    size              input                output\n");
    while(n){
        //当前层索引
        params.index = count; 
        fprintf(stderr, "%4d ", 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 == LOCAL){
            l = parse_local(options, params);
        }else if(lt == ACTIVE){
            l = parse_activation(options, params);
        }else if(lt == RNN){
            l = parse_rnn(options, params);
        }else if(lt == GRU){
            l = parse_gru(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 == REORG_OLD) {
            l = parse_reorg_old(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\n", s->type);
        }
        l.onlyforward = option_find_int_quiet(options, "onlyforward", 0);
        l.stopbackward = option_find_int_quiet(options, "stopbackward", 0);
        l.dontload = option_find_int_quiet(options, "dontload", 0);
        l.dontloadscales = option_find_int_quiet(options, "dontloadscales", 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){
            //记录当前层的信息,为下一层做准备
            params.h = l.out_h; 
            params.w = l.out_w;
            params.c = l.out_c;
            params.inputs = l.outputs;
        }
        if (l.bflops > 0) bflops += l.bflops;
    }   
    free_list(sections);
    net.outputs = get_network_output_size(net);
    net.output = get_network_output(net);
    printf("Total BFLOPS %5.3f \n", bflops);
    if(workspace_size){
        //printf("%ld\n", workspace_size);
#ifdef GPU
        if(gpu_index >= 0){
            net.workspace = cuda_make_array(0, workspace_size/sizeof(float) + 1);
        }else {
            net.workspace = calloc(1, workspace_size);
        }
#else
        net.workspace = calloc(1, workspace_size);
#endif
    }
    LAYER_TYPE lt = net.layers[net.n - 1].type;
    if ((net.w % 32 != 0 || net.h % 32 != 0) && (lt == YOLO || lt == REGION || lt == DETECTION)) {
        printf("\n Warning: width=%d and height=%d in cfg-file must be divisible by 32 for default networks Yolo v1/v2/v3!!! \n\n",
            net.w, net.h);
    }
    return net;
}