darknet/examples/nightmare.c

419 lines
13 KiB
C
Raw Normal View History

2017-06-02 06:31:13 +03:00
#include "darknet.h"
2015-07-08 10:36:43 +03:00
2017-06-08 23:47:31 +03:00
#include <math.h>
2016-03-01 00:54:12 +03:00
// ./darknet nightmare cfg/extractor.recon.cfg ~/trained/yolo-coco.conv frame6.png -reconstruct -iters 500 -i 3 -lambda .1 -rate .01 -smooth 2
2015-07-08 10:36:43 +03:00
float abs_mean(float *x, int n)
{
int i;
float sum = 0;
for (i = 0; i < n; ++i){
2016-02-01 02:52:03 +03:00
sum += fabs(x[i]);
2015-07-08 10:36:43 +03:00
}
return sum/n;
}
void calculate_loss(float *output, float *delta, int n, float thresh)
{
int i;
float mean = mean_array(output, n);
float var = variance_array(output, n);
for(i = 0; i < n; ++i){
if(delta[i] > mean + thresh*sqrt(var)) delta[i] = output[i];
else delta[i] = 0;
}
}
2015-11-14 23:34:17 +03:00
void optimize_picture(network *net, image orig, int max_layer, float scale, float rate, float thresh, int norm)
2015-07-08 10:36:43 +03:00
{
2016-03-01 00:54:12 +03:00
//scale_image(orig, 2);
//translate_image(orig, -1);
2015-07-08 10:36:43 +03:00
net->n = max_layer + 1;
int dx = rand()%16 - 8;
int dy = rand()%16 - 8;
int flip = rand()%2;
image crop = crop_image(orig, dx, dy, orig.w, orig.h);
image im = resize_image(crop, (int)(orig.w * scale), (int)(orig.h * scale));
if(flip) flip_image(im);
resize_network(net, im.w, im.h);
layer last = net->layers[net->n-1];
//net->layers[net->n - 1].activation = LINEAR;
image delta = make_image(im.w, im.h, im.c);
#ifdef GPU
net->delta_gpu = cuda_make_array(delta.data, im.w*im.h*im.c);
2017-11-26 23:27:21 +03:00
copy_cpu(net->inputs, im.data, 1, net->input, 1);
2015-07-08 10:36:43 +03:00
forward_network_gpu(net);
2017-06-18 23:05:37 +03:00
copy_gpu(last.outputs, last.output_gpu, 1, last.delta_gpu, 1);
2015-07-08 10:36:43 +03:00
cuda_pull_array(last.delta_gpu, last.delta, last.outputs);
calculate_loss(last.delta, last.delta, last.outputs, thresh);
cuda_push_array(last.delta_gpu, last.delta, last.outputs);
backward_network_gpu(net);
2015-07-08 10:36:43 +03:00
cuda_pull_array(net->delta_gpu, delta.data, im.w*im.h*im.c);
cuda_free(net->delta_gpu);
net->delta_gpu = 0;
2015-07-08 10:36:43 +03:00
#else
2017-11-26 23:27:21 +03:00
printf("\nnet: %d %d %d im: %d %d %d\n", net->w, net->h, net->inputs, im.w, im.h, im.c);
2017-11-21 22:34:46 +03:00
copy_cpu(net->inputs, im.data, 1, net->input, 1);
net->delta = delta.data;
forward_network(net);
2015-07-08 10:36:43 +03:00
copy_cpu(last.outputs, last.output, 1, last.delta, 1);
calculate_loss(last.output, last.delta, last.outputs, thresh);
backward_network(net);
2015-07-08 10:36:43 +03:00
#endif
if(flip) flip_image(delta);
//normalize_array(delta.data, delta.w*delta.h*delta.c);
image resized = resize_image(delta, orig.w, orig.h);
image out = crop_image(resized, -dx, -dy, orig.w, orig.h);
/*
image g = grayscale_image(out);
free_image(out);
out = g;
*/
//rate = rate / abs_mean(out.data, out.w*out.h*out.c);
image gray = make_image(out.w, out.h, out.c);
fill_image(gray, .5);
axpy_cpu(orig.w*orig.h*orig.c, -1, orig.data, 1, gray.data, 1);
axpy_cpu(orig.w*orig.h*orig.c, .1, gray.data, 1, out.data, 1);
2015-07-08 10:36:43 +03:00
2015-11-14 23:34:17 +03:00
if(norm) normalize_array(out.data, out.w*out.h*out.c);
2015-07-08 10:36:43 +03:00
axpy_cpu(orig.w*orig.h*orig.c, rate, out.data, 1, orig.data, 1);
/*
normalize_array(orig.data, orig.w*orig.h*orig.c);
scale_image(orig, sqrt(var));
translate_image(orig, mean);
*/
2016-03-01 00:54:12 +03:00
//translate_image(orig, 1);
//scale_image(orig, .5);
2015-07-08 10:36:43 +03:00
//normalize_image(orig);
constrain_image(orig);
free_image(crop);
free_image(im);
free_image(delta);
free_image(resized);
free_image(out);
}
2016-01-19 02:40:14 +03:00
void smooth(image recon, image update, float lambda, int num)
{
int i, j, k;
int ii, jj;
for(k = 0; k < recon.c; ++k){
for(j = 0; j < recon.h; ++j){
for(i = 0; i < recon.w; ++i){
int out_index = i + recon.w*(j + recon.h*k);
for(jj = j-num; jj <= j + num && jj < recon.h; ++jj){
if (jj < 0) continue;
for(ii = i-num; ii <= i + num && ii < recon.w; ++ii){
if (ii < 0) continue;
int in_index = ii + recon.w*(jj + recon.h*k);
update.data[out_index] += lambda * (recon.data[in_index] - recon.data[out_index]);
}
}
}
}
}
}
void reconstruct_picture(network *net, float *features, image recon, image update, float rate, float momentum, float lambda, int smooth_size, int iters)
2016-01-19 02:40:14 +03:00
{
2016-03-01 00:54:12 +03:00
int iter = 0;
for (iter = 0; iter < iters; ++iter) {
image delta = make_image(recon.w, recon.h, recon.c);
2016-01-19 02:40:14 +03:00
#ifdef GPU
2017-06-08 23:47:31 +03:00
layer l = get_network_output_layer(net);
cuda_push_array(net->input_gpu, recon.data, recon.w*recon.h*recon.c);
//cuda_push_array(net->truth_gpu, features, net->truths);
net->delta_gpu = cuda_make_array(delta.data, delta.w*delta.h*delta.c);
2016-01-19 02:40:14 +03:00
forward_network_gpu(net);
2017-04-13 00:22:53 +03:00
cuda_push_array(l.delta_gpu, features, l.outputs);
2017-06-18 23:05:37 +03:00
axpy_gpu(l.outputs, -1, l.output_gpu, 1, l.delta_gpu, 1);
backward_network_gpu(net);
2016-01-19 02:40:14 +03:00
cuda_pull_array(net->delta_gpu, delta.data, delta.w*delta.h*delta.c);
2016-01-19 02:40:14 +03:00
cuda_free(net->delta_gpu);
2016-01-19 02:40:14 +03:00
#else
net->input = recon.data;
net->delta = delta.data;
net->truth = features;
2016-01-19 02:40:14 +03:00
forward_network(net);
backward_network(net);
2016-01-19 02:40:14 +03:00
#endif
2017-04-13 00:22:53 +03:00
//normalize_array(delta.data, delta.w*delta.h*delta.c);
2016-03-01 00:54:12 +03:00
axpy_cpu(recon.w*recon.h*recon.c, 1, delta.data, 1, update.data, 1);
2017-04-13 00:22:53 +03:00
//smooth(recon, update, lambda, smooth_size);
2016-01-19 02:40:14 +03:00
2016-03-01 00:54:12 +03:00
axpy_cpu(recon.w*recon.h*recon.c, rate, update.data, 1, recon.data, 1);
scal_cpu(recon.w*recon.h*recon.c, momentum, update.data, 1);
2016-01-28 23:30:38 +03:00
2017-04-13 00:22:53 +03:00
float mag = mag_array(delta.data, recon.w*recon.h*recon.c);
printf("mag: %f\n", mag);
2016-03-01 00:54:12 +03:00
//scal_cpu(recon.w*recon.h*recon.c, 600/mag, recon.data, 1);
2016-01-28 23:30:38 +03:00
2016-03-01 00:54:12 +03:00
constrain_image(recon);
free_image(delta);
}
2016-01-19 02:40:14 +03:00
}
/*
void run_lsd(int argc, char **argv)
{
srand(0);
if(argc < 3){
fprintf(stderr, "usage: %s %s [cfg] [weights] [image] [options! (optional)]\n", argv[0], argv[1]);
return;
}
char *cfg = argv[2];
char *weights = argv[3];
char *input = argv[4];
int norm = find_int_arg(argc, argv, "-norm", 1);
int rounds = find_int_arg(argc, argv, "-rounds", 1);
int iters = find_int_arg(argc, argv, "-iters", 10);
float rate = find_float_arg(argc, argv, "-rate", .04);
float momentum = find_float_arg(argc, argv, "-momentum", .9);
float lambda = find_float_arg(argc, argv, "-lambda", .01);
char *prefix = find_char_arg(argc, argv, "-prefix", 0);
int reconstruct = find_arg(argc, argv, "-reconstruct");
int smooth_size = find_int_arg(argc, argv, "-smooth", 1);
network net = parse_network_cfg(cfg);
load_weights(&net, weights);
char *cfgbase = basecfg(cfg);
char *imbase = basecfg(input);
set_batch_network(&net, 1);
image im = load_image_color(input, 0, 0);
float *features = 0;
image update;
if (reconstruct){
im = letterbox_image(im, net->w, net->h);
int zz = 0;
network_predict(net, im.data);
image out_im = get_network_image(net);
image crop = crop_image(out_im, zz, zz, out_im.w-2*zz, out_im.h-2*zz);
//flip_image(crop);
image f_im = resize_image(crop, out_im.w, out_im.h);
free_image(crop);
printf("%d features\n", out_im.w*out_im.h*out_im.c);
im = resize_image(im, im.w, im.h);
f_im = resize_image(f_im, f_im.w, f_im.h);
features = f_im.data;
int i;
for(i = 0; i < 14*14*512; ++i){
features[i] += rand_uniform(-.19, .19);
}
free_image(im);
im = make_random_image(im.w, im.h, im.c);
update = make_image(im.w, im.h, im.c);
}
int e;
int n;
for(e = 0; e < rounds; ++e){
fprintf(stderr, "Iteration: ");
fflush(stderr);
for(n = 0; n < iters; ++n){
fprintf(stderr, "%d, ", n);
fflush(stderr);
if(reconstruct){
reconstruct_picture(net, features, im, update, rate, momentum, lambda, smooth_size, 1);
//if ((n+1)%30 == 0) rate *= .5;
show_image(im, "reconstruction");
#ifdef OPENCV
cvWaitKey(10);
#endif
}else{
int layer = max_layer + rand()%range - range/2;
int octave = rand()%octaves;
optimize_picture(&net, im, layer, 1/pow(1.33333333, octave), rate, thresh, norm);
}
}
fprintf(stderr, "done\n");
char buff[256];
if (prefix){
sprintf(buff, "%s/%s_%s_%d_%06d",prefix, imbase, cfgbase, max_layer, e);
}else{
sprintf(buff, "%s_%s_%d_%06d",imbase, cfgbase, max_layer, e);
}
printf("%d %s\n", e, buff);
save_image(im, buff);
//show_image(im, buff);
//cvWaitKey(0);
if(rotate){
image rot = rotate_image(im, rotate);
free_image(im);
im = rot;
}
image crop = crop_image(im, im.w * (1. - zoom)/2., im.h * (1.-zoom)/2., im.w*zoom, im.h*zoom);
image resized = resize_image(crop, im.w, im.h);
free_image(im);
free_image(crop);
im = resized;
}
}
*/
2015-07-08 10:36:43 +03:00
void run_nightmare(int argc, char **argv)
{
srand(0);
if(argc < 4){
fprintf(stderr, "usage: %s %s [cfg] [weights] [image] [layer] [options! (optional)]\n", argv[0], argv[1]);
return;
}
char *cfg = argv[2];
char *weights = argv[3];
char *input = argv[4];
int max_layer = atoi(argv[5]);
int range = find_int_arg(argc, argv, "-range", 1);
2015-11-14 23:34:17 +03:00
int norm = find_int_arg(argc, argv, "-norm", 1);
2015-07-08 10:36:43 +03:00
int rounds = find_int_arg(argc, argv, "-rounds", 1);
int iters = find_int_arg(argc, argv, "-iters", 10);
int octaves = find_int_arg(argc, argv, "-octaves", 4);
float zoom = find_float_arg(argc, argv, "-zoom", 1.);
float rate = find_float_arg(argc, argv, "-rate", .04);
float thresh = find_float_arg(argc, argv, "-thresh", 1.);
float rotate = find_float_arg(argc, argv, "-rotate", 0);
2016-01-19 02:40:14 +03:00
float momentum = find_float_arg(argc, argv, "-momentum", .9);
float lambda = find_float_arg(argc, argv, "-lambda", .01);
2015-07-10 01:22:14 +03:00
char *prefix = find_char_arg(argc, argv, "-prefix", 0);
2016-01-19 02:40:14 +03:00
int reconstruct = find_arg(argc, argv, "-reconstruct");
int smooth_size = find_int_arg(argc, argv, "-smooth", 1);
2015-07-08 10:36:43 +03:00
2017-11-26 23:27:21 +03:00
network *net = load_network(cfg, weights, 0);
2015-07-08 10:36:43 +03:00
char *cfgbase = basecfg(cfg);
char *imbase = basecfg(input);
set_batch_network(net, 1);
2015-07-08 10:36:43 +03:00
image im = load_image_color(input, 0, 0);
if(0){
float scale = 1;
if(im.w > 512 || im.h > 512){
if(im.w > im.h) scale = 512.0/im.w;
else scale = 512.0/im.h;
}
image resized = resize_image(im, scale*im.w, scale*im.h);
free_image(im);
im = resized;
}
//im = letterbox_image(im, net->w, net->h);
2015-07-08 10:36:43 +03:00
2016-03-01 00:54:12 +03:00
float *features = 0;
2016-01-19 02:40:14 +03:00
image update;
if (reconstruct){
net->n = max_layer;
im = letterbox_image(im, net->w, net->h);
2017-04-13 00:22:53 +03:00
//resize_network(&net, im.w, im.h);
2016-01-28 23:30:38 +03:00
network_predict(net, im.data);
if(net->layers[net->n-1].type == REGION){
2017-04-13 00:22:53 +03:00
printf("region!\n");
zero_objectness(net->layers[net->n-1]);
2017-04-13 00:22:53 +03:00
}
image out_im = copy_image(get_network_image(net));
/*
image crop = crop_image(out_im, zz, zz, out_im.w-2*zz, out_im.h-2*zz);
2016-01-28 23:30:38 +03:00
//flip_image(crop);
image f_im = resize_image(crop, out_im.w, out_im.h);
free_image(crop);
2017-04-13 00:22:53 +03:00
*/
2016-01-28 23:30:38 +03:00
printf("%d features\n", out_im.w*out_im.h*out_im.c);
2017-04-13 00:22:53 +03:00
features = out_im.data;
2016-01-28 23:30:38 +03:00
2017-04-13 00:22:53 +03:00
/*
2016-03-01 00:54:12 +03:00
int i;
2017-04-13 00:22:53 +03:00
for(i = 0; i < 14*14*512; ++i){
//features[i] += rand_uniform(-.19, .19);
2016-03-01 00:54:12 +03:00
}
2016-01-19 02:40:14 +03:00
free_image(im);
im = make_random_image(im.w, im.h, im.c);
2017-04-13 00:22:53 +03:00
*/
2016-01-19 02:40:14 +03:00
update = make_image(im.w, im.h, im.c);
}
2015-07-08 10:36:43 +03:00
int e;
int n;
for(e = 0; e < rounds; ++e){
2016-01-19 02:40:14 +03:00
fprintf(stderr, "Iteration: ");
fflush(stderr);
2015-07-08 10:36:43 +03:00
for(n = 0; n < iters; ++n){
fprintf(stderr, "%d, ", n);
fflush(stderr);
2016-01-19 02:40:14 +03:00
if(reconstruct){
2016-03-01 00:54:12 +03:00
reconstruct_picture(net, features, im, update, rate, momentum, lambda, smooth_size, 1);
//if ((n+1)%30 == 0) rate *= .5;
2016-01-19 02:40:14 +03:00
show_image(im, "reconstruction");
2016-03-01 00:54:12 +03:00
#ifdef OPENCV
2016-01-19 02:40:14 +03:00
cvWaitKey(10);
2016-03-01 00:54:12 +03:00
#endif
2016-01-19 02:40:14 +03:00
}else{
int layer = max_layer + rand()%range - range/2;
int octave = rand()%octaves;
optimize_picture(net, im, layer, 1/pow(1.33333333, octave), rate, thresh, norm);
2016-01-19 02:40:14 +03:00
}
2015-07-08 10:36:43 +03:00
}
fprintf(stderr, "done\n");
if(0){
image g = grayscale_image(im);
free_image(im);
im = g;
}
char buff[256];
2015-07-10 01:22:14 +03:00
if (prefix){
sprintf(buff, "%s/%s_%s_%d_%06d",prefix, imbase, cfgbase, max_layer, e);
}else{
sprintf(buff, "%s_%s_%d_%06d",imbase, cfgbase, max_layer, e);
}
2015-07-08 10:36:43 +03:00
printf("%d %s\n", e, buff);
save_image(im, buff);
//show_image(im, buff);
//cvWaitKey(0);
if(rotate){
image rot = rotate_image(im, rotate);
free_image(im);
im = rot;
}
image crop = crop_image(im, im.w * (1. - zoom)/2., im.h * (1.-zoom)/2., im.w*zoom, im.h*zoom);
image resized = resize_image(crop, im.w, im.h);
free_image(im);
free_image(crop);
im = resized;
}
}