blame | history | patch | commit | commitdiff | ignore whitespace
Joseph Redmon
2016-06-25 cf32e7e9b843560eb7ec3ed16e5b19f0f7156724 
 src/coco.c


@@ -1,50 +1,30 @@


#include <stdio.h>




#include "network.h"


#include "region_layer.h"


#include "detection_layer.h"


#include "cost_layer.h"


#include "utils.h"


#include "parser.h"


#include "box.h"


#include "demo.h"




#ifdef OPENCV


#include "opencv2/highgui/highgui_c.h"


#endif




void convert_detections(float *predictions, int classes, int num, int square, int side, int w, int h, float thresh, float **probs, box *boxes, int only_objectness);




char *coco_classes[] = {"person","bicycle","car","motorcycle","airplane","bus","train","truck","boat","traffic light","fire hydrant","stop sign","parking meter","bench","bird","cat","dog","horse","sheep","cow","elephant","bear","zebra","giraffe","backpack","umbrella","handbag","tie","suitcase","frisbee","skis","snowboard","sports ball","kite","baseball bat","baseball glove","skateboard","surfboard","tennis racket","bottle","wine glass","cup","fork","knife","spoon","bowl","banana","apple","sandwich","orange","broccoli","carrot","hot dog","pizza","donut","cake","chair","couch","potted plant","bed","dining table","toilet","tv","laptop","mouse","remote","keyboard","cell phone","microwave","oven","toaster","sink","refrigerator","book","clock","vase","scissors","teddy bear","hair drier","toothbrush"};




int coco_ids[] = {1,2,3,4,5,6,7,8,9,10,11,13,14,15,16,17,18,19,20,21,22,23,24,25,27,28,31,32,33,34,35,36,37,38,39,40,41,42,43,44,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,67,70,72,73,74,75,76,77,78,79,80,81,82,84,85,86,87,88,89,90};




void draw_coco(image im, int num, float thresh, box *boxes, float **probs, char *label)


{


    int classes = 80;


    int i;




    for(i = 0; i < num; ++i){


        int class = max_index(probs[i], classes);


        float prob = probs[i][class];


        if(prob > thresh){


            int width = sqrt(prob)*5 + 1;


            printf("%f %s\n", prob, coco_classes[class]);


            float red = get_color(0,class,classes);


            float green = get_color(1,class,classes);


            float blue = get_color(2,class,classes);


            box b = boxes[i];




            int left  = (b.x-b.w/2.)*im.w;


            int right = (b.x+b.w/2.)*im.w;


            int top   = (b.y-b.h/2.)*im.h;


            int bot   = (b.y+b.h/2.)*im.h;


            draw_box_width(im, left, top, right, bot, width, red, green, blue);


        }


    }


    show_image(im, label);


}


image coco_labels[80];




void train_coco(char *cfgfile, char *weightfile)


{


    //char *train_images = "/home/pjreddie/data/voc/test/train.txt";


    char *train_images = "/home/pjreddie/data/coco/train.txt";


    //char *train_images = "/home/pjreddie/data/coco/train.txt";


    char *train_images = "data/coco.trainval.txt";


    char *backup_directory = "/home/pjreddie/backup/";


    srand(time(0));


    data_seed = time(0);


@@ -121,34 +101,6 @@


    save_weights(net, buff);


}




void convert_coco_detections(float *predictions, int classes, int num, int square, int side, int w, int h, float thresh, float **probs, box *boxes, int only_objectness)


{


    int i,j,n;


    //int per_cell = 5*num+classes;


    for (i = 0; i < side*side; ++i){


        int row = i / side;


        int col = i % side;


        for(n = 0; n < num; ++n){


            int index = i*num + n;


            int p_index = side*side*classes + i*num + n;


            float scale = predictions[p_index];


            int box_index = side*side*(classes + num) + (i*num + n)*4;


            boxes[index].x = (predictions[box_index + 0] + col) / side * w;


            boxes[index].y = (predictions[box_index + 1] + row) / side * h;


            boxes[index].w = pow(predictions[box_index + 2], (square?2:1)) * w;


            boxes[index].h = pow(predictions[box_index + 3], (square?2:1)) * h;


            for(j = 0; j < classes; ++j){


                int class_index = i*classes;


                float prob = scale*predictions[class_index+j];


                probs[index][j] = (prob > thresh) ? prob : 0;


            }


            if(only_objectness){


                probs[index][0] = scale;


            }


        }


    }


}




void print_cocos(FILE *fp, int image_id, box *boxes, float **probs, int num_boxes, int classes, int w, int h)


{


    int i, j;


@@ -215,7 +167,7 @@


    int i=0;


    int t;




    float thresh = .001;


    float thresh = .01;


    int nms = 1;


    float iou_thresh = .5;




@@ -258,7 +210,7 @@


            float *predictions = network_predict(net, X);


            int w = val[t].w;


            int h = val[t].h;


            convert_coco_detections(predictions, classes, l.n, square, side, w, h, thresh, probs, boxes, 0);


            convert_detections(predictions, classes, l.n, square, side, w, h, thresh, probs, boxes, 0);


            if (nms) do_nms_sort(boxes, probs, side*side*l.n, classes, iou_thresh);


            print_cocos(fp, image_id, boxes, probs, side*side*l.n, classes, w, h);


            free_image(val[t]);


@@ -321,7 +273,7 @@


        image sized = resize_image(orig, net.w, net.h);


        char *id = basecfg(path);


        float *predictions = network_predict(net, sized.data);


        convert_coco_detections(predictions, classes, l.n, square, side, 1, 1, thresh, probs, boxes, 1);


        convert_detections(predictions, classes, l.n, square, side, 1, 1, thresh, probs, boxes, 1);


        if (nms) do_nms(boxes, probs, side*side*l.n, 1, nms_thresh);




        char *labelpath = find_replace(path, "images", "labels");


@@ -366,9 +318,10 @@


    if(weightfile){


        load_weights(&net, weightfile);


    }


    region_layer l = net.layers[net.n-1];


    detection_layer l = net.layers[net.n-1];


    set_batch_network(&net, 1);


    srand(2222222);


    float nms = .4;


    clock_t time;


    char buff[256];


    char *input = buff;


@@ -392,8 +345,10 @@


        time=clock();


        float *predictions = network_predict(net, X);


        printf("%s: Predicted in %f seconds.\n", input, sec(clock()-time));


        convert_coco_detections(predictions, l.classes, l.n, l.sqrt, l.side, 1, 1, thresh, probs, boxes, 0);


        draw_coco(im, l.side*l.side*l.n, thresh, boxes, probs, "predictions");


        convert_detections(predictions, l.classes, l.n, l.sqrt, l.side, 1, 1, thresh, probs, boxes, 0);


        if (nms) do_nms_sort(boxes, probs, l.side*l.side*l.n, l.classes, nms);


        draw_detections(im, l.side*l.side*l.n, thresh, boxes, probs, coco_classes, coco_labels, 80);


        show_image(im, "predictions");




        show_image(sized, "resized");


        free_image(im);


@@ -408,7 +363,16 @@




void run_coco(int argc, char **argv)


{


    int i;


    for(i = 0; i < 80; ++i){


        char buff[256];


        sprintf(buff, "data/labels/%s.png", coco_classes[i]);


        coco_labels[i] = load_image_color(buff, 0, 0);


    }


    float thresh = find_float_arg(argc, argv, "-thresh", .2);


    int cam_index = find_int_arg(argc, argv, "-c", 0);


    int frame_skip = find_int_arg(argc, argv, "-s", 0);




    if(argc < 4){


        fprintf(stderr, "usage: %s %s [train/test/valid] [cfg] [weights (optional)]\n", argv[0], argv[1]);


        return;


@@ -421,4 +385,5 @@


    else if(0==strcmp(argv[2], "train")) train_coco(cfg, weights);


    else if(0==strcmp(argv[2], "valid")) validate_coco(cfg, weights);


    else if(0==strcmp(argv[2], "recall")) validate_coco_recall(cfg, weights);


    else if(0==strcmp(argv[2], "demo")) demo(cfg, weights, thresh, cam_index, filename, coco_classes, coco_labels, 80, frame_skip);


}