Hacks to get nightmare to not break gridsizing

Joseph Redmon

2015-08-11 d0b9326a352ed2fbc3ae66fdef40b4533a2f211d

Hacks to get nightmare to not break gridsizing

 src/coco.c

@@ -1,3 +1,5 @@
#include <stdio.h>

#include "network.h"
#include "detection_layer.h"
#include "cost_layer.h"
@@ -8,6 +10,8 @@

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, float *box, int side, int objectness, char *label)
{
    int classes = 80;
@@ -144,7 +148,7 @@
    }
}

void print_cocos(FILE **fps, char *id, box *boxes, float **probs, int num_boxes, int classes, int w, int h)
void print_cocos(FILE *fp, int image_id, box *boxes, float **probs, int num_boxes, int classes, int w, int h)
{
    int i, j;
    for(i = 0; i < num_boxes*num_boxes; ++i){
@@ -158,13 +162,23 @@
        if (xmax > w) xmax = w;
        if (ymax > h) ymax = h;

        float bx = xmin;
        float by = ymin;
        float bw = xmax - xmin;
        float bh = ymax - ymin;

        for(j = 0; j < classes; ++j){
            if (probs[i][j]) fprintf(fps[j], "%s %f %f %f %f %f\n", id, probs[i][j],
                    xmin, ymin, xmax, ymax);
            if (probs[i][j]) fprintf(fp, "{\"image_id\":%d, \"category_id\":%d, \"bbox\":[%f, %f, %f, %f], \"score\":%f},\n", image_id, coco_ids[j], bx, by, bw, bh, probs[i][j]);
        }
    }
}

int get_coco_image_id(char *filename)
{
    char *p = strrchr(filename, '_');
    return atoi(p+1);
}

void validate_coco(char *cfgfile, char *weightfile)
{
    network net = parse_network_cfg(cfgfile);
@@ -176,8 +190,8 @@
    fprintf(stderr, "Learning Rate: %g, Momentum: %g, Decay: %g\n", net.learning_rate, net.momentum, net.decay);
    srand(time(0));

    char *base = "results/comp4_det_test_";
    list *plist = get_paths("data/voc.2012test.list");
    char *base = "/home/pjreddie/backup/";
    list *plist = get_paths("data/coco_val_5k.list");
    char **paths = (char **)list_to_array(plist);

    int classes = layer.classes;
@@ -186,12 +200,11 @@
    int num_boxes = sqrt(get_detection_layer_locations(layer));

    int j;
    FILE **fps = calloc(classes, sizeof(FILE *));
    for(j = 0; j < classes; ++j){
        char buff[1024];
        snprintf(buff, 1024, "%s%s.txt", base, coco_classes[j]);
        fps[j] = fopen(buff, "w");
    }
    snprintf(buff, 1024, "%s/coco_results.json", base);
    FILE *fp = fopen(buff, "w");
    fprintf(fp, "[\n");

    box *boxes = calloc(num_boxes*num_boxes, sizeof(box));
    float **probs = calloc(num_boxes*num_boxes, sizeof(float *));
    for(j = 0; j < num_boxes*num_boxes; ++j) probs[j] = calloc(classes, sizeof(float *));
@@ -200,7 +213,7 @@
    int i=0;
    int t;

    float thresh = .001;
    float thresh = .01;
    int nms = 1;
    float iou_thresh = .5;

@@ -226,19 +239,21 @@
        }
        for(t = 0; t < nthreads && i+t-nthreads < m; ++t){
            char *path = paths[i+t-nthreads];
            char *id = basecfg(path);
            int image_id = get_coco_image_id(path);
            float *X = val_resized[t].data;
            float *predictions = network_predict(net, X);
            int w = val[t].w;
            int h = val[t].h;
            convert_cocos(predictions, classes, objectness, background, num_boxes, w, h, thresh, probs, boxes);
            if (nms) do_nms(boxes, probs, num_boxes, classes, iou_thresh);
            print_cocos(fps, id, boxes, probs, num_boxes, classes, w, h);
            free(id);
            print_cocos(fp, image_id, boxes, probs, num_boxes, classes, w, h);
            free_image(val[t]);
            free_image(val_resized[t]);
        }
    }
    fseek(fp, -2, SEEK_CUR); 
    fprintf(fp, "\n]\n");
    fclose(fp);
    fprintf(stderr, "Total Detection Time: %f Seconds\n", (double)(time(0) - start));
}


 src/crop_layer_kernels.cu

@@ -5,8 +5,6 @@
#include "image.h"
}

#define BLOCK 256

__device__ float get_pixel_kernel(float *image, int w, int h, int x, int y, int c)
{
    if(x < 0 || x >= w || y < 0 || y >= h) return 0;

 src/cuda.h

@@ -5,7 +5,7 @@

#ifdef GPU

#define BLOCK 256
#define BLOCK 512

#include "cuda_runtime.h"
#include "curand.h"

 src/detection.c

@@ -79,7 +79,7 @@
    paths = (char **)list_to_array(plist);
    pthread_t load_thread = load_data_detection_thread(imgs, paths, plist->size, classes, net.w, net.h, side, side, background, &buffer);
    clock_t time;
    while(i*imgs < N*120){
    while(i*imgs < N*130){
        i += 1;
        time=clock();
        pthread_join(load_thread, 0);
@@ -95,7 +95,7 @@

        printf("%d: %f, %f avg, %lf seconds, %d images\n", i, loss, avg_loss, sec(clock()-time), i*imgs);
        if((i-1)*imgs <= N && i*imgs > N){
            fprintf(stderr, "Starting second stage...\n");
            fprintf(stderr, "First stage done\n");
            net.learning_rate *= 10;
            char buff[256];
            sprintf(buff, "%s/%s_first_stage.weights", backup_directory, base);
@@ -109,6 +109,13 @@
            save_weights(net, buff);
            return;
        }
        if((i-1)*imgs <= 120*N && i*imgs > N*120){
            fprintf(stderr, "Third stage done.\n");
            char buff[256];
            sprintf(buff, "%s/%s_third_stage.weights", backup_directory, base);
            net.layers[net.n-1].rescore = 1;
            save_weights(net, buff);
        }
        if(i%1000==0){
            char buff[256];
            sprintf(buff, "%s/%s_%d.weights", backup_directory, base, i);
@@ -176,7 +183,7 @@
    srand(time(0));

    char *base = "results/comp4_det_test_";
    list *plist = get_paths("data/voc.2012test.list");
    list *plist = get_paths("/home/pjreddie/data/voc/test/2007_test.txt");
    char **paths = (char **)list_to_array(plist);

    int classes = layer.classes;

 src/softmax_layer_kernels.cu

@@ -4,8 +4,6 @@
#include "blas.h"
}

#define BLOCK 256

__global__ void forward_softmax_layer_kernel(int n, int batch, float *input, float *output)
{
    int b = (blockIdx.x + blockIdx.y*gridDim.x) * blockDim.x + threadIdx.x;