2017-06-02 06:31:13 +03:00
|
|
|
#include "darknet.h"
|
2015-06-09 21:17:46 +03:00
|
|
|
|
|
|
|
void train_writing(char *cfgfile, char *weightfile)
|
|
|
|
{
|
2015-09-23 01:22:50 +03:00
|
|
|
char *backup_directory = "/home/pjreddie/backup/";
|
2015-06-09 21:17:46 +03:00
|
|
|
srand(time(0));
|
|
|
|
float avg_loss = -1;
|
|
|
|
char *base = basecfg(cfgfile);
|
|
|
|
printf("%s\n", base);
|
|
|
|
network net = parse_network_cfg(cfgfile);
|
|
|
|
if(weightfile){
|
|
|
|
load_weights(&net, weightfile);
|
|
|
|
}
|
|
|
|
printf("Learning Rate: %g, Momentum: %g, Decay: %g\n", net.learning_rate, net.momentum, net.decay);
|
2015-09-23 02:36:57 +03:00
|
|
|
int imgs = net.batch*net.subdivisions;
|
2015-06-09 21:17:46 +03:00
|
|
|
list *plist = get_paths("figures.list");
|
|
|
|
char **paths = (char **)list_to_array(plist);
|
|
|
|
clock_t time;
|
2015-09-23 02:36:57 +03:00
|
|
|
int N = plist->size;
|
|
|
|
printf("N: %d\n", N);
|
2015-09-24 00:13:43 +03:00
|
|
|
image out = get_network_image(net);
|
2015-09-23 02:36:57 +03:00
|
|
|
|
|
|
|
data train, buffer;
|
|
|
|
|
|
|
|
load_args args = {0};
|
|
|
|
args.w = net.w;
|
|
|
|
args.h = net.h;
|
2015-09-24 00:13:43 +03:00
|
|
|
args.out_w = out.w;
|
|
|
|
args.out_h = out.h;
|
2015-09-23 02:36:57 +03:00
|
|
|
args.paths = paths;
|
|
|
|
args.n = imgs;
|
|
|
|
args.m = N;
|
|
|
|
args.d = &buffer;
|
|
|
|
args.type = WRITING_DATA;
|
|
|
|
|
|
|
|
pthread_t load_thread = load_data_in_thread(args);
|
|
|
|
int epoch = (*net.seen)/N;
|
|
|
|
while(get_current_batch(net) < net.max_batches || net.max_batches == 0){
|
2015-06-09 21:17:46 +03:00
|
|
|
time=clock();
|
2015-09-23 02:36:57 +03:00
|
|
|
pthread_join(load_thread, 0);
|
|
|
|
train = buffer;
|
|
|
|
load_thread = load_data_in_thread(args);
|
2015-09-23 01:22:50 +03:00
|
|
|
printf("Loaded %lf seconds\n",sec(clock()-time));
|
2015-09-23 02:36:57 +03:00
|
|
|
|
2015-09-23 01:22:50 +03:00
|
|
|
time=clock();
|
2015-06-09 21:17:46 +03:00
|
|
|
float loss = train_network(net, train);
|
|
|
|
|
2015-09-23 01:22:50 +03:00
|
|
|
/*
|
2015-09-24 00:13:43 +03:00
|
|
|
image pred = float_to_image(64, 64, 1, out);
|
|
|
|
print_image(pred);
|
|
|
|
*/
|
2015-06-09 21:17:46 +03:00
|
|
|
|
2015-09-23 01:22:50 +03:00
|
|
|
/*
|
|
|
|
image im = float_to_image(256, 256, 3, train.X.vals[0]);
|
|
|
|
image lab = float_to_image(64, 64, 1, train.y.vals[0]);
|
|
|
|
image pred = float_to_image(64, 64, 1, out);
|
|
|
|
show_image(im, "image");
|
|
|
|
show_image(lab, "label");
|
|
|
|
print_image(lab);
|
|
|
|
show_image(pred, "pred");
|
|
|
|
cvWaitKey(0);
|
|
|
|
*/
|
|
|
|
|
2015-06-09 21:17:46 +03:00
|
|
|
if(avg_loss == -1) avg_loss = loss;
|
|
|
|
avg_loss = avg_loss*.9 + loss*.1;
|
2017-06-13 02:19:08 +03:00
|
|
|
printf("%ld, %.3f: %f, %f avg, %f rate, %lf seconds, %ld images\n", get_current_batch(net), (float)(*net.seen)/N, loss, avg_loss, get_current_rate(net), sec(clock()-time), *net.seen);
|
2015-09-24 00:13:43 +03:00
|
|
|
free_data(train);
|
|
|
|
if(get_current_batch(net)%100 == 0){
|
|
|
|
char buff[256];
|
2017-06-13 02:19:08 +03:00
|
|
|
sprintf(buff, "%s/%s_batch_%ld.weights", backup_directory, base, get_current_batch(net));
|
2015-09-24 00:13:43 +03:00
|
|
|
save_weights(net, buff);
|
|
|
|
}
|
|
|
|
if(*net.seen/N > epoch){
|
|
|
|
epoch = *net.seen/N;
|
|
|
|
char buff[256];
|
|
|
|
sprintf(buff, "%s/%s_%d.weights",backup_directory,base, epoch);
|
|
|
|
save_weights(net, buff);
|
|
|
|
}
|
2015-06-09 21:17:46 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-09-24 00:13:43 +03:00
|
|
|
void test_writing(char *cfgfile, char *weightfile, char *filename)
|
2015-09-23 01:22:50 +03:00
|
|
|
{
|
|
|
|
network net = parse_network_cfg(cfgfile);
|
|
|
|
if(weightfile){
|
|
|
|
load_weights(&net, weightfile);
|
|
|
|
}
|
|
|
|
set_batch_network(&net, 1);
|
|
|
|
srand(2222222);
|
|
|
|
clock_t time;
|
2015-09-24 00:13:43 +03:00
|
|
|
char buff[256];
|
|
|
|
char *input = buff;
|
|
|
|
while(1){
|
|
|
|
if(filename){
|
|
|
|
strncpy(input, filename, 256);
|
|
|
|
}else{
|
|
|
|
printf("Enter Image Path: ");
|
|
|
|
fflush(stdout);
|
|
|
|
input = fgets(input, 256, stdin);
|
|
|
|
if(!input) return;
|
|
|
|
strtok(input, "\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
image im = load_image_color(input, 0, 0);
|
|
|
|
resize_network(&net, im.w, im.h);
|
|
|
|
printf("%d %d %d\n", im.h, im.w, im.c);
|
|
|
|
float *X = im.data;
|
|
|
|
time=clock();
|
|
|
|
network_predict(net, X);
|
|
|
|
printf("%s: Predicted in %f seconds.\n", input, sec(clock()-time));
|
|
|
|
image pred = get_network_image(net);
|
|
|
|
|
|
|
|
image upsampled = resize_image(pred, im.w, im.h);
|
|
|
|
image thresh = threshold_image(upsampled, .5);
|
|
|
|
pred = thresh;
|
|
|
|
|
2015-09-23 01:22:50 +03:00
|
|
|
show_image(pred, "prediction");
|
2015-09-24 00:13:43 +03:00
|
|
|
show_image(im, "orig");
|
2015-09-23 01:22:50 +03:00
|
|
|
#ifdef OPENCV
|
2015-09-24 00:13:43 +03:00
|
|
|
cvWaitKey(0);
|
|
|
|
cvDestroyAllWindows();
|
2015-09-23 01:22:50 +03:00
|
|
|
#endif
|
|
|
|
|
2015-09-24 00:13:43 +03:00
|
|
|
free_image(upsampled);
|
|
|
|
free_image(thresh);
|
|
|
|
free_image(im);
|
|
|
|
if (filename) break;
|
|
|
|
}
|
2015-09-23 01:22:50 +03:00
|
|
|
}
|
|
|
|
|
2015-06-09 21:17:46 +03:00
|
|
|
void run_writing(int argc, char **argv)
|
|
|
|
{
|
2015-09-24 00:13:43 +03:00
|
|
|
if(argc < 4){
|
|
|
|
fprintf(stderr, "usage: %s %s [train/test/valid] [cfg] [weights (optional)]\n", argv[0], argv[1]);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
char *cfg = argv[3];
|
|
|
|
char *weights = (argc > 4) ? argv[4] : 0;
|
|
|
|
char *filename = (argc > 5) ? argv[5] : 0;
|
|
|
|
if(0==strcmp(argv[2], "train")) train_writing(cfg, weights);
|
|
|
|
else if(0==strcmp(argv[2], "test")) test_writing(cfg, weights, filename);
|
2015-06-09 21:17:46 +03:00
|
|
|
}
|
|
|
|
|