mirror/darknet
1
0
Fork 0
mirror of https://github.com/pjreddie/darknet.git synced 2023-08-10 21:13:14 +03:00
Code Issues Projects Releases Wiki Activity

CLEAN UP CLEAN UP EVERYBODY DO YOUR oh wait it's just me

Browse Source
This commit is contained in:
Joseph Redmon 2018-03-24 18:23:04 -07:00
parent e31c50127e
commit 777b098232
16 changed files with 620 additions and 337 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
Makefile
View File

@ -1,6 +1,6 @@
GPU=1
CUDNN=1
OPENCV=0
OPENCV=1
OPENMP=1
DEBUG=0
@ -26,7 +26,7 @@ ARFLAGS=rcs
OPTS=-Ofast
LDFLAGS= -lm -pthread
COMMON= -Iinclude/ -Isrc/
CFLAGS=-Wall -Wno-unknown-pragmas -Wfatal-errors -fPIC
CFLAGS=-Wall -Wno-unused-result -Wno-unknown-pragmas -Wfatal-errors -fPIC
ifeq ($(OPENMP), 1)
CFLAGS+= -fopenmp
@ -57,7 +57,7 @@ CFLAGS+= -DCUDNN
LDFLAGS+= -lcudnn
endif
OBJ=gemm.o utils.o cuda.o deconvolutional_layer.o convolutional_layer.o list.o image.o activations.o im2col.o col2im.o blas.o crop_layer.o dropout_layer.o maxpool_layer.o softmax_layer.o data.o matrix.o network.o connected_layer.o cost_layer.o parser.o option_list.o detection_layer.o route_layer.o upsample_layer.o box.o normalization_layer.o avgpool_layer.o layer.o local_layer.o shortcut_layer.o logistic_layer.o activation_layer.o rnn_layer.o gru_layer.o crnn_layer.o demo.o batchnorm_layer.o region_layer.o reorg_layer.o tree.o lstm_layer.o l2norm_layer.o
OBJ=gemm.o utils.o cuda.o deconvolutional_layer.o convolutional_layer.o list.o image.o activations.o im2col.o col2im.o blas.o crop_layer.o dropout_layer.o maxpool_layer.o softmax_layer.o data.o matrix.o network.o connected_layer.o cost_layer.o parser.o option_list.o detection_layer.o route_layer.o upsample_layer.o box.o normalization_layer.o avgpool_layer.o layer.o local_layer.o shortcut_layer.o logistic_layer.o activation_layer.o rnn_layer.o gru_layer.o crnn_layer.o demo.o batchnorm_layer.o region_layer.o reorg_layer.o tree.o lstm_layer.o l2norm_layer.o yolo_layer.o
EXECOBJA=captcha.o lsd.o super.o art.o tag.o cifar.o go.o rnn.o segmenter.o regressor.o classifier.o coco.o yolo.o detector.o nightmare.o darknet.o
ifeq ($(GPU), 1)
LDFLAGS+= -lstdc++

4
examples/classifier.c
View File

@ -51,7 +51,7 @@ void train_classifier(char *datacfg, char *cfgfile, char *weightfile, int *gpus,
if (tree) net->hierarchy = read_tree(tree);
int classes = option_find_int(options, "classes", 2);
char **labels;
char **labels = 0;
if(!tag){
labels = get_labels(label_list);
}
@ -161,7 +161,7 @@ void train_classifier(char *datacfg, char *cfgfile, char *weightfile, int *gpus,
pthread_join(load_thread, 0);
free_network(net);
free_ptrs((void**)labels, classes);
if(labels) free_ptrs((void**)labels, classes);
free_ptrs((void**)paths, plist->size);
free_list(plist);
free(base);

17
examples/coco.c
View File

@ -146,8 +146,6 @@ void validate_coco(char *cfg, char *weights)
FILE *fp = fopen(buff, "w");
fprintf(fp, "[\n");
detection *dets = make_network_boxes(net, 0);
int m = plist->size;
int i=0;
int t;
@ -195,9 +193,11 @@ void validate_coco(char *cfg, char *weights)
network_predict(net, X);
int w = val[t].w;
int h = val[t].h;
fill_network_boxes(net, w, h, thresh, 0, 0, 0, dets);
int nboxes = 0;
detection *dets = get_network_boxes(net, w, h, thresh, 0, 0, 0, &nboxes);
if (nms) do_nms_sort(dets, l.side*l.side*l.n, classes, iou_thresh);
print_cocos(fp, image_id, dets, l.side*l.side*l.n, classes, w, h);
free_detections(dets, nboxes);
free_image(val[t]);
free_image(val_resized[t]);
}
@ -231,7 +231,6 @@ void validate_coco_recall(char *cfgfile, char *weightfile)
snprintf(buff, 1024, "%s%s.txt", base, coco_classes[j]);
fps[j] = fopen(buff, "w");
}
detection *dets = make_network_boxes(net, 0);
int m = plist->size;
int i=0;
@ -252,7 +251,8 @@ void validate_coco_recall(char *cfgfile, char *weightfile)
char *id = basecfg(path);
network_predict(net, sized.data);
fill_network_boxes(net, orig.w, orig.h, thresh, 0, 0, 1, dets);
int nboxes = 0;
detection *dets = get_network_boxes(net, orig.w, orig.h, thresh, 0, 0, 1, &nboxes);
if (nms) do_nms_obj(dets, side*side*l.n, 1, nms);
char labelpath[4096];
@ -283,7 +283,7 @@ void validate_coco_recall(char *cfgfile, char *weightfile)
++correct;
}
}
free_detections(dets, nboxes);
fprintf(stderr, "%5d %5d %5d\tRPs/Img: %.2f\tIOU: %.2f%%\tRecall:%.2f%%\n", i, correct, total, (float)proposals/(i+1), avg_iou*100/total, 100.*correct/total);
free(id);
free_image(orig);
@ -302,7 +302,6 @@ void test_coco(char *cfgfile, char *weightfile, char *filename, float thresh)
clock_t time;
char buff[256];
char *input = buff;
detection *dets = make_network_boxes(net, 0);
while(1){
if(filename){
strncpy(input, filename, 256);
@ -320,12 +319,14 @@ void test_coco(char *cfgfile, char *weightfile, char *filename, float thresh)
network_predict(net, X);
printf("%s: Predicted in %f seconds.\n", input, sec(clock()-time));
fill_network_boxes(net, 1, 1, thresh, 0, 0, 0, dets);
int nboxes = 0;
detection *dets = get_network_boxes(net, im.w, im.h, thresh, 0, 0, 0, &nboxes);
if (nms) do_nms_sort(dets, l.side*l.side*l.n, l.classes, nms);
draw_detections(im, dets, l.side*l.side*l.n, thresh, coco_classes, alphabet, 80);
save_image(im, "prediction");
show_image(im, "predictions");
free_detections(dets, nboxes);
free_image(im);
free_image(sized);
#ifdef OPENCV

34
examples/detector.c
View File

@ -156,7 +156,9 @@ void train_detector(char *datacfg, char *cfgfile, char *weightfile, int *gpus, i
static int get_coco_image_id(char *filename)
{
char *p = strrchr(filename, '_');
char *p = strrchr(filename, '/');
char *c = strrchr(filename, '_');
if(c) p = c;
return atoi(p+1);
}
@ -467,6 +469,7 @@ void validate_detector(char *datacfg, char *cfgfile, char *weightfile, char *out
} else {
print_detector_detections(fps, id, dets, nboxes, classes, w, h);
}
free_detections(dets, nboxes);
free(id);
free_image(val[t]);
free_image(val_resized[t]);
@ -622,14 +625,13 @@ void test_detector(char *datacfg, char *cfgfile, char *weightfile, char *filenam
}
}
/*
void censor_detector(char *datacfg, char *cfgfile, char *weightfile, int cam_index, const char *filename, int class, float thresh, int skip)
{
#ifdef OPENCV
image **alphabet = load_alphabet();
char *base = basecfg(cfgfile);
network *net = load_network(cfgfile, weightfile, 0);
set_batch_network(net, 1);
list *options = read_data_cfg(datacfg);
srand(2222222);
CvCapture * cap;
@ -650,20 +652,11 @@ void censor_detector(char *datacfg, char *cfgfile, char *weightfile, int cam_ind
cvSetCaptureProperty(cap, CV_CAP_PROP_FRAME_HEIGHT, h);
}
int top = option_find_int(options, "top", 1);
char *label_list = option_find_str(options, "labels", 0);
char *name_list = option_find_str(options, "names", label_list);
char **names = get_labels(name_list);
int *indexes = calloc(top, sizeof(int));
if(!cap) error("Couldn't connect to webcam.\n");
cvNamedWindow(base, CV_WINDOW_NORMAL);
cvResizeWindow(base, 512, 512);
float fps = 0;
int i;
int count = 0;
float nms = .45;
while(1){
@ -709,11 +702,9 @@ void censor_detector(char *datacfg, char *cfgfile, char *weightfile, int cam_ind
void extract_detector(char *datacfg, char *cfgfile, char *weightfile, int cam_index, const char *filename, int class, float thresh, int skip)
{
#ifdef OPENCV
image **alphabet = load_alphabet();
char *base = basecfg(cfgfile);
network *net = load_network(cfgfile, weightfile, 0);
set_batch_network(net, 1);
list *options = read_data_cfg(datacfg);
srand(2222222);
CvCapture * cap;
@ -734,14 +725,6 @@ void extract_detector(char *datacfg, char *cfgfile, char *weightfile, int cam_in
cvSetCaptureProperty(cap, CV_CAP_PROP_FRAME_HEIGHT, h);
}
int top = option_find_int(options, "top", 1);
char *label_list = option_find_str(options, "labels", 0);
char *name_list = option_find_str(options, "names", label_list);
char **names = get_labels(name_list);
int *indexes = calloc(top, sizeof(int));
if(!cap) error("Couldn't connect to webcam.\n");
cvNamedWindow(base, CV_WINDOW_NORMAL);
cvResizeWindow(base, 512, 512);
@ -795,6 +778,7 @@ void extract_detector(char *datacfg, char *cfgfile, char *weightfile, int cam_in
}
#endif
}
*/
/*
void network_detect(network *net, image im, float thresh, float hier_thresh, float nms, detection *dets)
@ -848,15 +832,13 @@ void run_detector(int argc, char **argv)
int width = find_int_arg(argc, argv, "-w", 0);
int height = find_int_arg(argc, argv, "-h", 0);
int fps = find_int_arg(argc, argv, "-fps", 0);
int class = find_int_arg(argc, argv, "-class", 0);
//int class = find_int_arg(argc, argv, "-class", 0);
char *datacfg = argv[3];
char *cfg = argv[4];
char *weights = (argc > 5) ? argv[5] : 0;
char *filename = (argc > 6) ? argv[6]: 0;
if(0==strcmp(argv[2], "test")) test_detector(datacfg, cfg, weights, filename, thresh, hier_thresh, outfile, fullscreen);
else if(0==strcmp(argv[2], "extract")) extract_detector(datacfg, cfg, weights, cam_index, filename, class, thresh, frame_skip);
else if(0==strcmp(argv[2], "censor")) censor_detector(datacfg, cfg, weights, cam_index, filename, class, thresh, frame_skip);
else if(0==strcmp(argv[2], "train")) train_detector(datacfg, cfg, weights, gpus, ngpus, clear);
else if(0==strcmp(argv[2], "valid")) validate_detector(datacfg, cfg, weights, outfile);
else if(0==strcmp(argv[2], "valid2")) validate_detector_flip(datacfg, cfg, weights, outfile);
@ -868,4 +850,6 @@ void run_detector(int argc, char **argv)
char **names = get_labels(name_list);
demo(cfg, weights, thresh, cam_index, filename, names, classes, frame_skip, prefix, avg, hier_thresh, width, height, fps, fullscreen);
}
//else if(0==strcmp(argv[2], "extract")) extract_detector(datacfg, cfg, weights, cam_index, filename, class, thresh, frame_skip);
//else if(0==strcmp(argv[2], "censor")) censor_detector(datacfg, cfg, weights, cam_index, filename, class, thresh, frame_skip);
}

7
examples/lsd.c
View File

@ -1,3 +1,4 @@
#include <math.h>
#include "darknet.h"
/*
@ -478,7 +479,7 @@ void test_dcgan(char *cfgfile, char *weightfile)
clock_t time;
char buff[256];
char *input = buff;
int i, imlayer = 0;
int imlayer = 0;
imlayer = net->n-1;
@ -615,7 +616,7 @@ void train_prog(char *cfg, char *weight, char *acfg, char *aweight, int clear, i
forward_network(anet);
backward_network(anet);
float genaloss = *anet->cost / anet->batch;
//float genaloss = *anet->cost / anet->batch;
scal_gpu(imlayer.outputs*imlayer.batch, 1, imerror, 1);
scal_gpu(imlayer.outputs*imlayer.batch, 0, gnet->layers[gnet->n-1].delta_gpu, 1);
@ -785,7 +786,7 @@ void train_dcgan(char *cfg, char *weight, char *acfg, char *aweight, int clear,
forward_network(anet);
backward_network(anet);
float genaloss = *anet->cost / anet->batch;
//float genaloss = *anet->cost / anet->batch;
//printf("%f\n", genaloss);
scal_gpu(imlayer.outputs*imlayer.batch, 1, imerror, 1);

4
examples/rnn.c
View File

@ -100,8 +100,8 @@ float_pair get_seq2seq_data(char **source, char **dest, int n, int characters, s
float *y = calloc(batch * steps * characters, sizeof(float));
for(i = 0; i < batch; ++i){
int index = rand()%n;
int slen = strlen(source[index]);
int dlen = strlen(dest[index]);
//int slen = strlen(source[index]);
//int dlen = strlen(dest[index]);
for(j = 0; j < steps; ++j){
unsigned char curr = source[index][j];
unsigned char next = dest[index][j];

16
examples/yolo.c
View File

@ -133,7 +133,6 @@ void validate_yolo(char *cfg, char *weights)
image *buf = calloc(nthreads, sizeof(image));
image *buf_resized = calloc(nthreads, sizeof(image));
pthread_t *thr = calloc(nthreads, sizeof(pthread_t));
detection *dets = make_network_boxes(net, 0);
load_args args = {0};
args.w = net->w;
@ -167,9 +166,11 @@ void validate_yolo(char *cfg, char *weights)
network_predict(net, X);
int w = val[t].w;
int h = val[t].h;
fill_network_boxes(net, w, h, thresh, 0, 0, 0, dets);
int nboxes = 0;
detection *dets = get_network_boxes(net, w, h, thresh, 0, 0, 0, &nboxes);
if (nms) do_nms_sort(dets, l.side*l.side*l.n, classes, iou_thresh);
print_yolo_detections(fps, id, l.side*l.side*l.n, classes, w, h, dets);
free_detections(dets, nboxes);
free(id);
free_image(val[t]);
free_image(val_resized[t]);
@ -200,7 +201,6 @@ void validate_yolo_recall(char *cfg, char *weights)
snprintf(buff, 1024, "%s%s.txt", base, voc_names[j]);
fps[j] = fopen(buff, "w");
}
detection *dets = make_network_boxes(net, 0);
int m = plist->size;
int i=0;
@ -221,7 +221,8 @@ void validate_yolo_recall(char *cfg, char *weights)
char *id = basecfg(path);
network_predict(net, sized.data);
fill_network_boxes(net, orig.w, orig.h, thresh, 0, 0, 1, dets);
int nboxes = 0;
detection *dets = get_network_boxes(net, orig.w, orig.h, thresh, 0, 0, 1, &nboxes);
if (nms) do_nms_obj(dets, side*side*l.n, 1, nms);
char labelpath[4096];
@ -254,6 +255,7 @@ void validate_yolo_recall(char *cfg, char *weights)
}
fprintf(stderr, "%5d %5d %5d\tRPs/Img: %.2f\tIOU: %.2f%%\tRecall:%.2f%%\n", i, correct, total, (float)proposals/(i+1), avg_iou*100/total, 100.*correct/total);
free_detections(dets, nboxes);
free(id);
free_image(orig);
free_image(sized);
@ -271,7 +273,6 @@ void test_yolo(char *cfgfile, char *weightfile, char *filename, float thresh)
char buff[256];
char *input = buff;
float nms=.4;
detection *dets = make_network_boxes(net, 0);
while(1){
if(filename){
strncpy(input, filename, 256);
@ -289,13 +290,14 @@ void test_yolo(char *cfgfile, char *weightfile, char *filename, float thresh)
network_predict(net, X);
printf("%s: Predicted in %f seconds.\n", input, sec(clock()-time));
fill_network_boxes(net, 1, 1, thresh, 0, 0, 0, dets);
int nboxes = 0;
detection *dets = get_network_boxes(net, im.w, im.h, thresh, 0, 0, 0, &nboxes);
if (nms) do_nms_sort(dets, l.side*l.side*l.n, l.classes, nms);
draw_detections(im, dets, l.side*l.side*l.n, thresh, voc_names, alphabet, 20);
save_image(im, "predictions");
show_image(im, "predictions");
free_detections(dets, nboxes);
free_image(im);
free_image(sized);
#ifdef OPENCV

5
include/darknet.h
View File

@ -85,6 +85,7 @@ typedef enum {
NETWORK,
XNOR,
REGION,
YOLO,
REORG,
UPSAMPLE,
LOGXENT,
@ -674,6 +675,7 @@ void get_detection_detections(layer l, int w, int h, float thresh, detection *de
char *option_find_str(list *l, char *key, char *def);
int option_find_int(list *l, char *key, int def);
int option_find_int_quiet(list *l, char *key, int def);
network *parse_network_cfg(char *filename);
void save_weights(network *net, char *filename);
@ -682,7 +684,8 @@ void save_weights_upto(network *net, char *filename, int cutoff);
void load_weights_upto(network *net, char *filename, int start, int cutoff);
void zero_objectness(layer l);
int get_region_detections(layer l, int w, int h, int netw, int neth, float thresh, int *map, float tree_thresh, int relative, detection *dets);
void get_region_detections(layer l, int w, int h, int netw, int neth, float thresh, int *map, float tree_thresh, int relative, detection *dets);
int get_yolo_detections(layer l, int w, int h, int netw, int neth, float thresh, int *map, int relative, detection *dets);
void free_network(network *net);
void set_batch_network(network *net, int b);
void set_temp_network(network *net, float t);

4
src/demo.c
View File

@ -50,7 +50,7 @@ void *detect_in_thread(void *ptr)
if(l.type == DETECTION){
get_detection_boxes(l, 1, 1, demo_thresh, probs, boxes, 0);
} else */
detection *dets;
detection *dets = 0;
int nboxes = 0;
if (l.type == REGION){
dets = get_network_boxes(net, buff[0].w, buff[0].h, demo_thresh, demo_hier, 0, 1, &nboxes);
@ -174,8 +174,6 @@ void demo(char *cfgfile, char *weightfile, float thresh, int cam_index, const ch
if(!cap) error("Couldn't connect to webcam.\n");
int i;
buff[0] = get_image_from_stream(cap);
buff[1] = copy_image(buff[0]);
buff[2] = copy_image(buff[0]);

19
src/network.c
View File

@ -17,6 +17,7 @@
#include "activation_layer.h"
#include "detection_layer.h"
#include "region_layer.h"
#include "yolo_layer.h"
#include "normalization_layer.h"
#include "batchnorm_layer.h"
#include "maxpool_layer.h"
@ -151,6 +152,8 @@ char *get_layer_string(LAYER_TYPE a)
return "detection";
case REGION:
return "region";
case YOLO:
return "yolo";
case DROPOUT:
return "dropout";
case CROP:
@ -376,6 +379,8 @@ int resize_network(network *net, int w, int h)
resize_maxpool_layer(&l, w, h);
}else if(l.type == REGION){
resize_region_layer(&l, w, h);
}else if(l.type == YOLO){
resize_yolo_layer(&l, w, h);
}else if(l.type == ROUTE){
resize_route_layer(&l, net);
}else if(l.type == SHORTCUT){
@ -508,10 +513,10 @@ int num_detections(network *net, float thresh)
int s = 0;
for(i = 0; i < net->n; ++i){
layer l = net->layers[i];
if(l.type == REGION){
s += region_num_detections(l, thresh);
if(l.type == YOLO){
s += yolo_num_detections(l, thresh);
}
if(l.type == DETECTION){
if(l.type == DETECTION || l.type == REGION){
s += l.w*l.h*l.n;
}
}
@ -539,10 +544,14 @@ void fill_network_boxes(network *net, int w, int h, float thresh, float hier, in
int j;
for(j = 0; j < net->n; ++j){
layer l = net->layers[j];
if(l.type == REGION){
int count = get_region_detections(l, w, h, net->w, net->h, thresh, map, hier, relative, dets);
if(l.type == YOLO){
int count = get_yolo_detections(l, w, h, net->w, net->h, thresh, map, relative, dets);
dets += count;
}
if(l.type == REGION){
get_region_detections(l, w, h, net->w, net->h, thresh, map, hier, relative, dets);
dets += l.w*l.h*l.n;
}
if(l.type == DETECTION){
get_detection_detections(l, w, h, thresh, dets);
dets += l.w*l.h*l.n;

1
src/option_list.h
View File

@ -12,7 +12,6 @@ typedef struct{
int read_option(char *s, list *options);
void option_insert(list *l, char *key, char *val);
char *option_find(list *l, char *key);
int option_find_int_quiet(list *l, char *key, int def);
float option_find_float(list *l, char *key, float def);
float option_find_float_quiet(list *l, char *key, float def);
void option_unused(list *l);

98
src/parser.c
View File

@ -26,6 +26,7 @@
#include "option_list.h"
#include "parser.h"
#include "region_layer.h"
#include "yolo_layer.h"
#include "reorg_layer.h"
#include "rnn_layer.h"
#include "route_layer.h"
@ -50,6 +51,7 @@ LAYER_TYPE string_to_layer_type(char * type)
if (strcmp(type, "[cost]")==0) return COST;
if (strcmp(type, "[detection]")==0) return DETECTION;
if (strcmp(type, "[region]")==0) return REGION;
if (strcmp(type, "[yolo]")==0) return YOLO;
if (strcmp(type, "[local]")==0) return LOCAL;
if (strcmp(type, "[conv]")==0
|| strcmp(type, "[convolutional]")==0) return CONVOLUTIONAL;
@ -277,14 +279,8 @@ softmax_layer parse_softmax(list *options, size_params params)
return layer;
}
layer parse_region(list *options, size_params params)
int *parse_yolo_mask(char *a, int *num)
{
int coords = option_find_int(options, "coords", 4);
int classes = option_find_int(options, "classes", 20);
int total = option_find_int(options, "num", 1);
int num = total;
char *a = option_find_str(options, "mask", 0);
int *mask = 0;
if(a){
int len = strlen(a);
@ -299,36 +295,29 @@ layer parse_region(list *options, size_params params)
mask[i] = val;
a = strchr(a, ',')+1;
}
num = n;
*num = n;
}
layer l = make_region_layer(params.batch, params.w, params.h, num, total, mask, classes, coords);
return mask;
}
layer parse_yolo(list *options, size_params params)
{
int classes = option_find_int(options, "classes", 20);
int total = option_find_int(options, "num", 1);
int num = total;
char *a = option_find_str(options, "mask", 0);
int *mask = parse_yolo_mask(a, &num);
layer l = make_yolo_layer(params.batch, params.w, params.h, num, total, mask, classes);
assert(l.outputs == params.inputs);
l.log = option_find_int_quiet(options, "log", 0);
l.sqrt = option_find_int_quiet(options, "sqrt", 0);
l.softmax = option_find_int(options, "softmax", 0);
l.background = option_find_int_quiet(options, "background", 0);
l.max_boxes = option_find_int_quiet(options, "max",90);
l.jitter = option_find_float(options, "jitter", .2);
l.rescore = option_find_int_quiet(options, "rescore",0);
l.ignore_thresh = option_find_float(options, "ignore_thresh", .5);
l.truth_thresh = option_find_float(options, "truth_thresh", 1);
l.classfix = option_find_int_quiet(options, "classfix", 0);
l.absolute = option_find_int_quiet(options, "absolute", 0);
l.random = option_find_int_quiet(options, "random", 0);
l.coord_scale = option_find_float(options, "coord_scale", 1);
l.object_scale = option_find_float(options, "object_scale", 1);
l.noobject_scale = option_find_float(options, "noobject_scale", 1);
l.mask_scale = option_find_float_quiet(options, "mask_scale", 1);
l.class_scale = option_find_float(options, "class_scale", 1);
l.bias_match = option_find_int_quiet(options, "bias_match",0);
l.focus = option_find_float_quiet(options, "focus", 0);
char *tree_file = option_find_str(options, "tree", 0);
if (tree_file) l.softmax_tree = read_tree(tree_file);
char *map_file = option_find_str(options, "map", 0);
if (map_file) l.map = read_map(map_file);
@ -348,6 +337,59 @@ layer parse_region(list *options, size_params params)
}
return l;
}
layer parse_region(list *options, size_params params)
{
int coords = option_find_int(options, "coords", 4);
int classes = option_find_int(options, "classes", 20);
int num = option_find_int(options, "num", 1);
layer l = make_region_layer(params.batch, params.w, params.h, num, classes, coords);
assert(l.outputs == params.inputs);
l.log = option_find_int_quiet(options, "log", 0);
l.sqrt = option_find_int_quiet(options, "sqrt", 0);
l.softmax = option_find_int(options, "softmax", 0);
l.background = option_find_int_quiet(options, "background", 0);
l.max_boxes = option_find_int_quiet(options, "max",30);
l.jitter = option_find_float(options, "jitter", .2);
l.rescore = option_find_int_quiet(options, "rescore",0);
l.thresh = option_find_float(options, "thresh", .5);
l.classfix = option_find_int_quiet(options, "classfix", 0);
l.absolute = option_find_int_quiet(options, "absolute", 0);
l.random = option_find_int_quiet(options, "random", 0);
l.coord_scale = option_find_float(options, "coord_scale", 1);
l.object_scale = option_find_float(options, "object_scale", 1);
l.noobject_scale = option_find_float(options, "noobject_scale", 1);
l.mask_scale = option_find_float(options, "mask_scale", 1);
l.class_scale = option_find_float(options, "class_scale", 1);
l.bias_match = option_find_int_quiet(options, "bias_match",0);
char *tree_file = option_find_str(options, "tree", 0);
if (tree_file) l.softmax_tree = read_tree(tree_file);
char *map_file = option_find_str(options, "map", 0);
if (map_file) l.map = read_map(map_file);
char *a = option_find_str(options, "anchors", 0);
if(a){
int len = strlen(a);
int n = 1;
int i;
for(i = 0; i < len; ++i){
if (a[i] == ',') ++n;
}
for(i = 0; i < n; ++i){
float bias = atof(a);
l.biases[i] = bias;
a = strchr(a, ',')+1;
}
}
return l;
}
detection_layer parse_detection(list *options, size_params params)
{
int coords = option_find_int(options, "coords", 1);
@ -747,6 +789,8 @@ network *parse_network_cfg(char *filename)
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){

258
src/region_layer.c
View File

@ -10,14 +10,12 @@
#include <string.h>
#include <stdlib.h>
layer make_region_layer(int batch, int w, int h, int n, int total, int *mask, int classes, int coords)
layer make_region_layer(int batch, int w, int h, int n, int classes, int coords)
{
int i;
layer l = {0};
l.type = REGION;
l.n = n;
l.total = total;
l.batch = batch;
l.h = h;
l.w = w;
@ -28,21 +26,15 @@ layer make_region_layer(int batch, int w, int h, int n, int total, int *mask, in
l.classes = classes;
l.coords = coords;
l.cost = calloc(1, sizeof(float));
l.biases = calloc(total*2, sizeof(float));
if(mask) l.mask = mask;
else{
l.mask = calloc(n, sizeof(int));
for(i = 0; i < n; ++i){
l.mask[i] = i;
}
}
l.biases = calloc(n*2, sizeof(float));
l.bias_updates = calloc(n*2, sizeof(float));
l.outputs = h*w*n*(classes + coords + 1);
l.inputs = l.outputs;
l.truths = 90*(l.coords + 1);
l.truths = 30*(l.coords + 1);
l.delta = calloc(batch*l.outputs, sizeof(float));
l.output = calloc(batch*l.outputs, sizeof(float));
for(i = 0; i < total*2; ++i){
int i;
for(i = 0; i < n*2; ++i){
l.biases[i] = .5;
}
@ -81,37 +73,30 @@ void resize_region_layer(layer *l, int w, int h)
#endif
}
box get_region_box(float *x, float *biases, int n, int index, int i, int j, int lw, int lh, int w, int h, int stride)
box get_region_box(float *x, float *biases, int n, int index, int i, int j, int w, int h, int stride)
{
box b;
b.x = (i + x[index + 0*stride]) / lw;
b.y = (j + x[index + 1*stride]) / lh;
b.x = (i + x[index + 0*stride]) / w;
b.y = (j + x[index + 1*stride]) / h;
b.w = exp(x[index + 2*stride]) * biases[2*n] / w;
b.h = exp(x[index + 3*stride]) * biases[2*n+1] / h;
return b;
}
float delta_region_box(box truth, float *x, float *biases, int n, int index, int i, int j, int lw, int lh, int w, int h, float *delta, float scale, int stride)
float delta_region_box(box truth, float *x, float *biases, int n, int index, int i, int j, int w, int h, float *delta, float scale, int stride)
{
box pred = get_region_box(x, biases, n, index, i, j, lw, lh, w, h, stride);
box pred = get_region_box(x, biases, n, index, i, j, w, h, stride);
float iou = box_iou(pred, truth);
float tx = (truth.x*lw - i);
float ty = (truth.y*lh - j);
float tx = (truth.x*w - i);
float ty = (truth.y*h - j);
float tw = log(truth.w*w / biases[2*n]);
float th = log(truth.h*h / biases[2*n + 1]);
//printf("%f %f %f %f\n", tx, ty, tw, th);
delta[index + 0*stride] = scale * (tx - x[index + 0*stride]);
delta[index + 1*stride] = scale * (ty - x[index + 1*stride]);
delta[index + 2*stride] = scale * (tw - x[index + 2*stride]);
delta[index + 3*stride] = scale * (th - x[index + 3*stride]);
//printf("x: %f %f\n",tx , x[index + 0*stride]);
//printf("y: %f %f\n",ty , x[index + 1*stride]);
//printf("w: %f %f\n",tw , x[index + 2*stride]);
//printf("h: %f %f\n\n",th , x[index + 3*stride]);
//printf("%f %f %f %f\n", x[index + 0*stride], x[index + 1*stride], x[index + 2*stride], x[index + 3*stride]);
return iou;
}
@ -124,7 +109,7 @@ void delta_region_mask(float *truth, float *x, int n, int index, float *delta, i
}
void delta_region_class(float *output, float *delta, int index, int class, int classes, tree *hier, float scale, int stride, float *avg_cat, int tag, float focus)
void delta_region_class(float *output, float *delta, int index, int class, int classes, tree *hier, float scale, int stride, float *avg_cat, int tag)
{
int i, n;
if(hier){
@ -140,30 +125,15 @@ void delta_region_class(float *output, float *delta, int index, int class, int c
class = hier->parent[class];
}
if(avg_cat) *avg_cat += pred;
*avg_cat += pred;
} else {
if (delta[index] && tag){
if(focus){
float y = -1;
float p = output[index + stride*class];
float lg = p > .0000000001 ? log(p) : -10;
delta[index + stride*class] = y * pow(1-p, focus) * (focus*p*lg + p - 1);
}else{
delta[index + stride*class] = scale * (1 - output[index + stride*class]);
if(avg_cat) *avg_cat += output[index + stride*class];
}
return;
}
for(n = 0; n < classes; ++n){
if(focus){
float y = (n == class) ? -1 : 1;
float p = (n == class) ? output[index + stride*n] : 1 - output[index + stride*n];
float lg = p > .0000000001 ? log(p) : -10;
delta[index + stride*n] = y * pow(1-p, focus) * (focus*p*lg + p - 1);
}else{
delta[index + stride*n] = scale * (((n == class)?1 : 0) - output[index + stride*n]);
}
if(n == class && avg_cat) *avg_cat += output[index + stride*n];
if(n == class) *avg_cat += output[index + stride*n];
}
}
}
@ -219,7 +189,6 @@ void forward_region_layer(const layer l, network net)
if(!net.train) return;
float avg_iou = 0;
float recall = 0;
float recall75 = 0;
float avg_cat = 0;
float avg_obj = 0;
float avg_anyobj = 0;
@ -229,7 +198,7 @@ void forward_region_layer(const layer l, network net)
for (b = 0; b < l.batch; ++b) {
if(l.softmax_tree){
int onlyclass = 0;
for(t = 0; t < l.max_boxes; ++t){
for(t = 0; t < 30; ++t){
box truth = float_to_box(net.truth + t*(l.coords + 1) + b*l.truths, 1);
if(!truth.x) break;
int class = net.truth[t*(l.coords + 1) + b*l.truths + l.coords];
@ -249,7 +218,7 @@ void forward_region_layer(const layer l, network net)
}
int class_index = entry_index(l, b, maxi, l.coords + 1);
int obj_index = entry_index(l, b, maxi, l.coords);
delta_region_class(l.output, l.delta, class_index, class, l.classes, l.softmax_tree, l.class_scale, l.w*l.h, &avg_cat, !l.softmax, l.focus);
delta_region_class(l.output, l.delta, class_index, class, l.classes, l.softmax_tree, l.class_scale, l.w*l.h, &avg_cat, !l.softmax);
if(l.output[obj_index] < .3) l.delta[obj_index] = l.object_scale * (.3 - l.output[obj_index]);
else l.delta[obj_index] = 0;
l.delta[obj_index] = 0;
@ -264,50 +233,36 @@ void forward_region_layer(const layer l, network net)
for (i = 0; i < l.w; ++i) {
for (n = 0; n < l.n; ++n) {
int box_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, 0);
box pred = get_region_box(l.output, l.biases, l.mask[n], box_index, i, j, l.w, l.h, net.w, net.h, l.w*l.h);
box pred = get_region_box(l.output, l.biases, n, box_index, i, j, l.w, l.h, l.w*l.h);
float best_iou = 0;
int best_t = 0;
for(t = 0; t < l.max_boxes; ++t){
for(t = 0; t < 30; ++t){
box truth = float_to_box(net.truth + t*(l.coords + 1) + b*l.truths, 1);
if(!truth.x) break;
float iou = box_iou(pred, truth);
if (iou > best_iou) {
best_iou = iou;
best_t = t;
}
}
int obj_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, l.coords);
avg_anyobj += l.output[obj_index];
l.delta[obj_index] = l.noobject_scale * (0 - l.output[obj_index]);
if(l.background) l.delta[obj_index] = l.noobject_scale * (1 - l.output[obj_index]);
if (best_iou > l.ignore_thresh) {
if (best_iou > l.thresh) {
l.delta[obj_index] = 0;
}
if (best_iou > l.truth_thresh) {
l.delta[obj_index] = l.object_scale * (1 - l.output[obj_index]);
int class = net.truth[best_t*(l.coords + 1) + b*l.truths + l.coords];
if (l.map) class = l.map[class];
int class_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, l.coords + 1);
delta_region_class(l.output, l.delta, class_index, class, l.classes, l.softmax_tree, l.class_scale, l.w*l.h, 0, !l.softmax, l.focus);
box truth = float_to_box(net.truth + best_t*(l.coords + 1) + b*l.truths, 1);
delta_region_box(truth, l.output, l.biases, l.mask[n], box_index, i, j, l.w, l.h, net.w, net.h, l.delta, l.coord_scale*(2-truth.w*truth.h), l.w*l.h);
}
/*
if(*(net.seen) < 12800){
box truth = {0};
truth.x = (i + .5)/l.w;
truth.y = (j + .5)/l.h;
truth.w = l.biases[2*l.mask[n]]/net.w;
truth.h = l.biases[2*l.mask[n]+1]/net.h;
delta_region_box(truth, l.output, l.biases, l.mask[n], box_index, i, j, l.w, l.h, net.w, net.h, l.delta, .01, l.w*l.h);
}
*/
truth.w = l.biases[2*n]/l.w;
truth.h = l.biases[2*n+1]/l.h;
delta_region_box(truth, l.output, l.biases, n, box_index, i, j, l.w, l.h, l.delta, .01, l.w*l.h);
}
}
}
for(t = 0; t < l.max_boxes; ++t){
}
for(t = 0; t < 30; ++t){
box truth = float_to_box(net.truth + t*(l.coords + 1) + b*l.truths, 1);
if(!truth.x) break;
@ -315,39 +270,35 @@ void forward_region_layer(const layer l, network net)
int best_n = 0;
i = (truth.x * l.w);
j = (truth.y * l.h);
//printf("%d %f %d %f\n", i, truth.x*l.w, j, truth.y*l.h);
box truth_shift = truth;
truth_shift.x = 0;
truth_shift.y = 0;
//printf("index %d %d\n",i, j);
for(n = 0; n < l.total; ++n){
box pred = {0};
pred.w = l.biases[2*n]/net.w;
pred.h = l.biases[2*n+1]/net.h;
//printf("pred: (%f, %f) %f x %f\n", pred.x, pred.y, pred.w, pred.h);
for(n = 0; n < l.n; ++n){
int box_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, 0);
box pred = get_region_box(l.output, l.biases, n, box_index, i, j, l.w, l.h, l.w*l.h);
if(l.bias_match){
pred.w = l.biases[2*n]/l.w;
pred.h = l.biases[2*n+1]/l.h;
}
pred.x = 0;
pred.y = 0;
float iou = box_iou(pred, truth_shift);
if (iou > best_iou){
best_iou = iou;
best_n = n;
}
}
//printf("%d %f (%f, %f) %f x %f\n", best_n, best_iou, truth.x, truth.y, truth.w, truth.h);
int mask_n = int_index(l.mask, best_n, l.n);
//printf("%d %d\n", best_n, mask_n);
if(mask_n >= 0){
int box_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, 0);
float iou = delta_region_box(truth, l.output, l.biases, best_n, box_index, i, j, l.w, l.h, net.w, net.h, l.delta, l.coord_scale*(2-truth.w*truth.h), l.w*l.h);
int box_index = entry_index(l, b, best_n*l.w*l.h + j*l.w + i, 0);
float iou = delta_region_box(truth, l.output, l.biases, best_n, box_index, i, j, l.w, l.h, l.delta, l.coord_scale * (2 - truth.w*truth.h), l.w*l.h);
if(l.coords > 4){
int mask_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, 4);
int mask_index = entry_index(l, b, best_n*l.w*l.h + j*l.w + i, 4);
delta_region_mask(net.truth + t*(l.coords + 1) + b*l.truths + 5, l.output, l.coords - 4, mask_index, l.delta, l.w*l.h, l.mask_scale);
}
if(iou > .5) recall += 1;
if(iou > .75) recall75 += 1;
avg_iou += iou;
//l.delta[best_index + 4] = iou - l.output[best_index + 4];
int obj_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, l.coords);
int obj_index = entry_index(l, b, best_n*l.w*l.h + j*l.w + i, l.coords);
avg_obj += l.output[obj_index];
l.delta[obj_index] = l.object_scale * (1 - l.output[obj_index]);
if (l.rescore) {
@ -359,16 +310,14 @@ void forward_region_layer(const layer l, network net)
int class = net.truth[t*(l.coords + 1) + b*l.truths + l.coords];
if (l.map) class = l.map[class];
int class_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, l.coords + 1);
delta_region_class(l.output, l.delta, class_index, class, l.classes, l.softmax_tree, l.class_scale, l.w*l.h, &avg_cat, !l.softmax, l.focus);
int class_index = entry_index(l, b, best_n*l.w*l.h + j*l.w + i, l.coords + 1);
delta_region_class(l.output, l.delta, class_index, class, l.classes, l.softmax_tree, l.class_scale, l.w*l.h, &avg_cat, !l.softmax);
++count;
++class_count;
}
}
}
//printf("\n");
*(l.cost) = pow(mag_array(l.delta, l.outputs * l.batch), 2);
printf("Region %d Avg IOU: %f, Class: %f, Obj: %f, No Obj: %f, .5R: %f, .75R: %f, count: %d\n", net.index, avg_iou/count, avg_cat/class_count, avg_obj/count, avg_anyobj/(l.w*l.h*l.n*l.batch), recall/count, recall75/count, count);
printf("Region Avg IOU: %f, Class: %f, Obj: %f, No Obj: %f, Avg Recall: %f, count: %d\n", avg_iou/count, avg_cat/class_count, avg_obj/count, avg_anyobj/(l.w*l.h*l.n*l.batch), recall/count, count);
}
void backward_region_layer(const layer l, network net)
@ -412,27 +361,11 @@ void correct_region_boxes(detection *dets, int n, int w, int h, int netw, int ne
}
}
int region_num_detections(layer l, float thresh)
{
int i, n;
int count = 0;
for (i = 0; i < l.w*l.h; ++i){
int row = i / l.w;
int col = i % l.w;
for(n = 0; n < l.n; ++n){
int index = n*l.w*l.h + i;
int obj_index = entry_index(l, 0, n*l.w*l.h + i, l.coords);
if(l.output[obj_index] > thresh){
++count;
}
}
}
return count;
}
void avg_flipped_region(layer l)
void get_region_detections(layer l, int w, int h, int netw, int neth, float thresh, int *map, float tree_thresh, int relative, detection *dets)
{
int i,j,n,z;
float *predictions = l.output;
if (l.batch == 2) {
float *flip = l.output + l.outputs;
for (j = 0; j < l.h; ++j) {
for (i = 0; i < l.w/2; ++i) {
@ -454,31 +387,21 @@ void avg_flipped_region(layer l)
for(i = 0; i < l.outputs; ++i){
l.output[i] = (l.output[i] + flip[i])/2.;
}
}
int get_region_detections(layer l, int w, int h, int netw, int neth, float thresh, int *map, float tree_thresh, int relative, detection *dets)
{
int i,j,n,z;
float *predictions = l.output;
if (l.batch == 2) avg_flipped_region(l);
int count = 0;
}
for (i = 0; i < l.w*l.h; ++i){
int row = i / l.w;
int col = i % l.w;
for(n = 0; n < l.n; ++n){
int obj_index = entry_index(l, 0, n*l.w*l.h + i, l.coords);
if(predictions[obj_index] <= thresh) continue;
int index = count;
++count;
int box_index = entry_index(l, 0, n*l.w*l.h + i, 0);
int mask_index = entry_index(l, 0, n*l.w*l.h + i, 4);
for (j = 0; j < l.classes; ++j) {
int index = n*l.w*l.h + i;
for(j = 0; j < l.classes; ++j){
dets[index].prob[j] = 0;
}
int obj_index = entry_index(l, 0, n*l.w*l.h + i, l.coords);
int box_index = entry_index(l, 0, n*l.w*l.h + i, 0);
int mask_index = entry_index(l, 0, n*l.w*l.h + i, 4);
float scale = l.background ? 1 : predictions[obj_index];
dets[index].bbox = get_region_box(predictions, l.biases, l.mask[n], box_index, col, row, l.w, l.h, netw, neth, l.w*l.h);
dets[index].bbox = get_region_box(predictions, l.biases, n, box_index, col, row, l.w, l.h, l.w*l.h);
dets[index].objectness = scale > thresh ? scale : 0;
dets[index].classes = l.classes;
if(dets[index].mask){
for(j = 0; j < l.coords - 4; ++j){
dets[index].mask[j] = l.output[mask_index + j*l.w*l.h];
@ -510,8 +433,7 @@ int get_region_detections(layer l, int w, int h, int netw, int neth, float thres
}
}
}
correct_region_boxes(dets, count, w, h, netw, neth, relative);
return count;
correct_region_boxes(dets, l.w*l.h*l.n, w, h, netw, neth, relative);
}
#ifdef GPU
@ -537,80 +459,8 @@ void forward_region_layer_gpu(const layer l, network net)
if (l.softmax_tree){
int index = entry_index(l, 0, 0, l.coords + 1);
softmax_tree(net.input_gpu + index, l.w*l.h, l.batch*l.n, l.inputs/l.n, 1, l.output_gpu + index, *l.softmax_tree);
/*
int mmin = 9000;
int mmax = 0;
int i;
for(i = 0; i < l.softmax_tree->groups; ++i){
int group_size = l.softmax_tree->group_size[i];
if (group_size < mmin) mmin = group_size;
if (group_size > mmax) mmax = group_size;
}
//printf("%d %d %d \n", l.softmax_tree->groups, mmin, mmax);
*/
/*
// TIMING CODE
int zz;
int number = 1000;
int count = 0;
int i;
for (i = 0; i < l.softmax_tree->groups; ++i) {
int group_size = l.softmax_tree->group_size[i];
count += group_size;
}
printf("%d %d\n", l.softmax_tree->groups, count);
{
double then = what_time_is_it_now();
for(zz = 0; zz < number; ++zz){
int index = entry_index(l, 0, 0, 5);
softmax_tree(net.input_gpu + index, l.w*l.h, l.batch*l.n, l.inputs/l.n, 1, l.output_gpu + index, *l.softmax_tree);
}
cudaDeviceSynchronize();
printf("Good GPU Timing: %f\n", what_time_is_it_now() - then);
}
{
double then = what_time_is_it_now();
for(zz = 0; zz < number; ++zz){
int i;
int count = 5;
for (i = 0; i < l.softmax_tree->groups; ++i) {
int group_size = l.softmax_tree->group_size[i];
int index = entry_index(l, 0, 0, count);
softmax_gpu(net.input_gpu + index, group_size, l.batch*l.n, l.inputs/l.n, l.w*l.h, 1, l.w*l.h, 1, l.output_gpu + index);
count += group_size;
}
}
cudaDeviceSynchronize();
printf("Bad GPU Timing: %f\n", what_time_is_it_now() - then);
}
{
double then = what_time_is_it_now();
for(zz = 0; zz < number; ++zz){
int i;
int count = 5;
for (i = 0; i < l.softmax_tree->groups; ++i) {
int group_size = l.softmax_tree->group_size[i];
softmax_cpu(net.input + count, group_size, l.batch, l.inputs, l.n*l.w*l.h, 1, l.n*l.w*l.h, l.temperature, l.output + count);
count += group_size;
}
}
cudaDeviceSynchronize();
printf("CPU Timing: %f\n", what_time_is_it_now() - then);
}
*/
/*
int i;
int count = 5;
for (i = 0; i < l.softmax_tree->groups; ++i) {
int group_size = l.softmax_tree->group_size[i];
int index = entry_index(l, 0, 0, count);
softmax_gpu(net.input_gpu + index, group_size, l.batch*l.n, l.inputs/l.n, l.w*l.h, 1, l.w*l.h, 1, l.output_gpu + index);
count += group_size;
}
*/
} else if (l.softmax) {
int index = entry_index(l, 0, 0, l.coords + !l.background);
//printf("%d\n", index);
softmax_gpu(net.input_gpu + index, l.classes + l.background, l.batch*l.n, l.inputs/l.n, l.w*l.h, 1, l.w*l.h, 1, l.output_gpu + index);
}
if(!net.train || l.onlyforward){
@ -631,13 +481,13 @@ void backward_region_layer_gpu(const layer l, network net)
for (b = 0; b < l.batch; ++b){
for(n = 0; n < l.n; ++n){
int index = entry_index(l, b, n*l.w*l.h, 0);
//gradient_array_gpu(l.output_gpu + index, 2*l.w*l.h, LOGISTIC, l.delta_gpu + index);
gradient_array_gpu(l.output_gpu + index, 2*l.w*l.h, LOGISTIC, l.delta_gpu + index);
if(l.coords > 4){
index = entry_index(l, b, n*l.w*l.h, 4);
gradient_array_gpu(l.output_gpu + index, (l.coords - 4)*l.w*l.h, LOGISTIC, l.delta_gpu + index);
}
index = entry_index(l, b, n*l.w*l.h, l.coords);
//if(!l.background) gradient_array_gpu(l.output_gpu + index, l.w*l.h, LOGISTIC, l.delta_gpu + index);
if(!l.background) gradient_array_gpu(l.output_gpu + index, l.w*l.h, LOGISTIC, l.delta_gpu + index);
}
}
axpy_gpu(l.batch*l.inputs, 1, l.delta_gpu, 1, net.delta_gpu, 1);

3
src/region_layer.h
View File

@ -5,11 +5,10 @@
#include "layer.h"
#include "network.h"
layer make_region_layer(int batch, int h, int w, int n, int total, int *mask, int classes, int coords);
layer make_region_layer(int batch, int w, int h, int n, int classes, int coords);
void forward_region_layer(const layer l, network net);
void backward_region_layer(const layer l, network net);
void resize_region_layer(layer *l, int w, int h);
int region_num_detections(layer l, float thresh);
#ifdef GPU
void forward_region_layer_gpu(const layer l, network net);

374
src/yolo_layer.c Normal file
View File

@ -0,0 +1,374 @@
#include "yolo_layer.h"
#include "activations.h"
#include "blas.h"
#include "box.h"
#include "cuda.h"
#include "utils.h"
#include <stdio.h>
#include <assert.h>
#include <string.h>
#include <stdlib.h>
layer make_yolo_layer(int batch, int w, int h, int n, int total, int *mask, int classes)
{
int i;
layer l = {0};
l.type = YOLO;
l.n = n;
l.total = total;
l.batch = batch;
l.h = h;
l.w = w;
l.c = n*(classes + 4 + 1);
l.out_w = l.w;
l.out_h = l.h;
l.out_c = l.c;
l.classes = classes;
l.cost = calloc(1, sizeof(float));
l.biases = calloc(total*2, sizeof(float));
if(mask) l.mask = mask;
else{
l.mask = calloc(n, sizeof(int));
for(i = 0; i < n; ++i){
l.mask[i] = i;
}
}
l.bias_updates = calloc(n*2, sizeof(float));
l.outputs = h*w*n*(classes + 4 + 1);
l.inputs = l.outputs;
l.truths = 90*(4 + 1);
l.delta = calloc(batch*l.outputs, sizeof(float));
l.output = calloc(batch*l.outputs, sizeof(float));
for(i = 0; i < total*2; ++i){
l.biases[i] = .5;
}
l.forward = forward_yolo_layer;
l.backward = backward_yolo_layer;
#ifdef GPU
l.forward_gpu = forward_yolo_layer_gpu;
l.backward_gpu = backward_yolo_layer_gpu;
l.output_gpu = cuda_make_array(l.output, batch*l.outputs);
l.delta_gpu = cuda_make_array(l.delta, batch*l.outputs);
#endif
fprintf(stderr, "detection\n");
srand(0);
return l;
}
void resize_yolo_layer(layer *l, int w, int h)
{
l->w = w;
l->h = h;
l->outputs = h*w*l->n*(l->classes + 4 + 1);
l->inputs = l->outputs;
l->output = realloc(l->output, l->batch*l->outputs*sizeof(float));
l->delta = realloc(l->delta, l->batch*l->outputs*sizeof(float));
#ifdef GPU
cuda_free(l->delta_gpu);
cuda_free(l->output_gpu);
l->delta_gpu = cuda_make_array(l->delta, l->batch*l->outputs);
l->output_gpu = cuda_make_array(l->output, l->batch*l->outputs);
#endif
}
box get_yolo_box(float *x, float *biases, int n, int index, int i, int j, int lw, int lh, int w, int h, int stride)
{
box b;
b.x = (i + x[index + 0*stride]) / lw;
b.y = (j + x[index + 1*stride]) / lh;
b.w = exp(x[index + 2*stride]) * biases[2*n] / w;
b.h = exp(x[index + 3*stride]) * biases[2*n+1] / h;
return b;
}
float delta_yolo_box(box truth, float *x, float *biases, int n, int index, int i, int j, int lw, int lh, int w, int h, float *delta, float scale, int stride)
{
box pred = get_yolo_box(x, biases, n, index, i, j, lw, lh, w, h, stride);
float iou = box_iou(pred, truth);
float tx = (truth.x*lw - i);
float ty = (truth.y*lh - j);
float tw = log(truth.w*w / biases[2*n]);
float th = log(truth.h*h / biases[2*n + 1]);
delta[index + 0*stride] = scale * (tx - x[index + 0*stride]);
delta[index + 1*stride] = scale * (ty - x[index + 1*stride]);
delta[index + 2*stride] = scale * (tw - x[index + 2*stride]);
delta[index + 3*stride] = scale * (th - x[index + 3*stride]);
return iou;
}
void delta_yolo_class(float *output, float *delta, int index, int class, int classes, int stride, float *avg_cat)
{
int n;
if (delta[index]){
delta[index + stride*class] = 1 - output[index + stride*class];
if(avg_cat) *avg_cat += output[index + stride*class];
return;
}
for(n = 0; n < classes; ++n){
delta[index + stride*n] = ((n == class)?1 : 0) - output[index + stride*n];
if(n == class && avg_cat) *avg_cat += output[index + stride*n];
}
}
static int entry_index(layer l, int batch, int location, int entry)
{
int n = location / (l.w*l.h);
int loc = location % (l.w*l.h);
return batch*l.outputs + n*l.w*l.h*(4+l.classes+1) + entry*l.w*l.h + loc;
}
void forward_yolo_layer(const layer l, network net)
{
int i,j,b,t,n;
memcpy(l.output, net.input, l.outputs*l.batch*sizeof(float));
#ifndef GPU
for (b = 0; b < l.batch; ++b){
for(n = 0; n < l.n; ++n){
int index = entry_index(l, b, n*l.w*l.h, 0);
activate_array(l.output + index, 2*l.w*l.h, LOGISTIC);
index = entry_index(l, b, n*l.w*l.h, 4);
activate_array(l.output + index, (1+l.classes)*l.w*l.h, LOGISTIC);
}
}
#endif
memset(l.delta, 0, l.outputs * l.batch * sizeof(float));
if(!net.train) return;
float avg_iou = 0;
float recall = 0;
float recall75 = 0;
float avg_cat = 0;
float avg_obj = 0;
float avg_anyobj = 0;
int count = 0;
int class_count = 0;
*(l.cost) = 0;
for (b = 0; b < l.batch; ++b) {
for (j = 0; j < l.h; ++j) {
for (i = 0; i < l.w; ++i) {
for (n = 0; n < l.n; ++n) {
int box_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, 0);
box pred = get_yolo_box(l.output, l.biases, l.mask[n], box_index, i, j, l.w, l.h, net.w, net.h, l.w*l.h);
float best_iou = 0;
int best_t = 0;
for(t = 0; t < l.max_boxes; ++t){
box truth = float_to_box(net.truth + t*(4 + 1) + b*l.truths, 1);
if(!truth.x) break;
float iou = box_iou(pred, truth);
if (iou > best_iou) {
best_iou = iou;
best_t = t;
}
}
int obj_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, 4);
avg_anyobj += l.output[obj_index];
l.delta[obj_index] = 0 - l.output[obj_index];
if (best_iou > l.ignore_thresh) {
l.delta[obj_index] = 0;
}
if (best_iou > l.truth_thresh) {
l.delta[obj_index] = 1 - l.output[obj_index];
int class = net.truth[best_t*(4 + 1) + b*l.truths + 4];
if (l.map) class = l.map[class];
int class_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, 4 + 1);
delta_yolo_class(l.output, l.delta, class_index, class, l.classes, l.w*l.h, 0);
box truth = float_to_box(net.truth + best_t*(4 + 1) + b*l.truths, 1);
delta_yolo_box(truth, l.output, l.biases, l.mask[n], box_index, i, j, l.w, l.h, net.w, net.h, l.delta, (2-truth.w*truth.h), l.w*l.h);
}
}
}
}
for(t = 0; t < l.max_boxes; ++t){
box truth = float_to_box(net.truth + t*(4 + 1) + b*l.truths, 1);
if(!truth.x) break;
float best_iou = 0;
int best_n = 0;
i = (truth.x * l.w);
j = (truth.y * l.h);
box truth_shift = truth;
truth_shift.x = truth_shift.y = 0;
for(n = 0; n < l.total; ++n){
box pred = {0};
pred.w = l.biases[2*n]/net.w;
pred.h = l.biases[2*n+1]/net.h;
float iou = box_iou(pred, truth_shift);
if (iou > best_iou){
best_iou = iou;
best_n = n;
}
}
int mask_n = int_index(l.mask, best_n, l.n);
if(mask_n >= 0){
int box_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, 0);
float iou = delta_yolo_box(truth, l.output, l.biases, best_n, box_index, i, j, l.w, l.h, net.w, net.h, l.delta, (2-truth.w*truth.h), l.w*l.h);
int obj_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, 4);
avg_obj += l.output[obj_index];
l.delta[obj_index] = 1 - l.output[obj_index];
int class = net.truth[t*(4 + 1) + b*l.truths + 4];
if (l.map) class = l.map[class];
int class_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, 4 + 1);
delta_yolo_class(l.output, l.delta, class_index, class, l.classes, l.w*l.h, &avg_cat);
++count;
++class_count;
if(iou > .5) recall += 1;
if(iou > .75) recall75 += 1;
avg_iou += iou;
}
}
}
*(l.cost) = pow(mag_array(l.delta, l.outputs * l.batch), 2);
printf("Region %d Avg IOU: %f, Class: %f, Obj: %f, No Obj: %f, .5R: %f, .75R: %f, count: %d\n", net.index, avg_iou/count, avg_cat/class_count, avg_obj/count, avg_anyobj/(l.w*l.h*l.n*l.batch), recall/count, recall75/count, count);
}
void backward_yolo_layer(const layer l, network net)
{
axpy_cpu(l.batch*l.inputs, 1, l.delta, 1, net.delta, 1);
}
void correct_yolo_boxes(detection *dets, int n, int w, int h, int netw, int neth, int relative)
{
int i;
int new_w=0;
int new_h=0;
if (((float)netw/w) < ((float)neth/h)) {
new_w = netw;
new_h = (h * netw)/w;
} else {
new_h = neth;
new_w = (w * neth)/h;
}
for (i = 0; i < n; ++i){
box b = dets[i].bbox;
b.x = (b.x - (netw - new_w)/2./netw) / ((float)new_w/netw);
b.y = (b.y - (neth - new_h)/2./neth) / ((float)new_h/neth);
b.w *= (float)netw/new_w;
b.h *= (float)neth/new_h;
if(!relative){
b.x *= w;
b.w *= w;
b.y *= h;
b.h *= h;
}
dets[i].bbox = b;
}
}
int yolo_num_detections(layer l, float thresh)
{
int i, n;
int count = 0;
for (i = 0; i < l.w*l.h; ++i){
for(n = 0; n < l.n; ++n){
int obj_index = entry_index(l, 0, n*l.w*l.h + i, 4);
if(l.output[obj_index] > thresh){
++count;
}
}
}
return count;
}
void avg_flipped_yolo(layer l)
{
int i,j,n,z;
float *flip = l.output + l.outputs;
for (j = 0; j < l.h; ++j) {
for (i = 0; i < l.w/2; ++i) {
for (n = 0; n < l.n; ++n) {
for(z = 0; z < l.classes + 4 + 1; ++z){
int i1 = z*l.w*l.h*l.n + n*l.w*l.h + j*l.w + i;
int i2 = z*l.w*l.h*l.n + n*l.w*l.h + j*l.w + (l.w - i - 1);
float swap = flip[i1];
flip[i1] = flip[i2];
flip[i2] = swap;
if(z == 0){
flip[i1] = -flip[i1];
flip[i2] = -flip[i2];
}
}
}
}
}
for(i = 0; i < l.outputs; ++i){
l.output[i] = (l.output[i] + flip[i])/2.;
}
}
int get_yolo_detections(layer l, int w, int h, int netw, int neth, float thresh, int *map, int relative, detection *dets)
{
int i,j,n;
float *predictions = l.output;
if (l.batch == 2) avg_flipped_yolo(l);
int count = 0;
for (i = 0; i < l.w*l.h; ++i){
int row = i / l.w;
int col = i % l.w;
for(n = 0; n < l.n; ++n){
int obj_index = entry_index(l, 0, n*l.w*l.h + i, 4);
float objectness = predictions[obj_index];
if(objectness <= thresh) continue;
int box_index = entry_index(l, 0, n*l.w*l.h + i, 0);
dets[count].bbox = get_yolo_box(predictions, l.biases, l.mask[n], box_index, col, row, l.w, l.h, netw, neth, l.w*l.h);
dets[count].objectness = objectness;
dets[count].classes = l.classes;
for(j = 0; j < l.classes; ++j){
int class_index = entry_index(l, 0, n*l.w*l.h + i, 4 + 1 + j);
float prob = objectness*predictions[class_index];
dets[count].prob[j] = (prob > thresh) ? prob : 0;
}
++count;
}
}
correct_yolo_boxes(dets, count, w, h, netw, neth, relative);
return count;
}
#ifdef GPU
void forward_yolo_layer_gpu(const layer l, network net)
{
copy_gpu(l.batch*l.inputs, net.input_gpu, 1, l.output_gpu, 1);
int b, n;
for (b = 0; b < l.batch; ++b){
for(n = 0; n < l.n; ++n){
int index = entry_index(l, b, n*l.w*l.h, 0);
activate_array_gpu(l.output_gpu + index, 2*l.w*l.h, LOGISTIC);
index = entry_index(l, b, n*l.w*l.h, 4);
activate_array_gpu(l.output_gpu + index, (1+l.classes)*l.w*l.h, LOGISTIC);
}
}
if(!net.train || l.onlyforward){
cuda_pull_array(l.output_gpu, l.output, l.batch*l.outputs);
return;
}
cuda_pull_array(l.output_gpu, net.input, l.batch*l.inputs);
forward_yolo_layer(l, net);
cuda_push_array(l.delta_gpu, l.delta, l.batch*l.outputs);
}
void backward_yolo_layer_gpu(const layer l, network net)
{
axpy_gpu(l.batch*l.inputs, 1, l.delta_gpu, 1, net.delta_gpu, 1);
}
#endif

19
src/yolo_layer.h Normal file
View File

@ -0,0 +1,19 @@
#ifndef YOLO_LAYER_H
#define YOLO_LAYER_H
#include "darknet.h"
#include "layer.h"
#include "network.h"
layer make_yolo_layer(int batch, int w, int h, int n, int total, int *mask, int classes);
void forward_yolo_layer(const layer l, network net);
void backward_yolo_layer(const layer l, network net);
void resize_yolo_layer(layer *l, int w, int h);
int yolo_num_detections(layer l, float thresh);
#ifdef GPU
void forward_yolo_layer_gpu(const layer l, network net);
void backward_yolo_layer_gpu(layer l, network net);
#endif
#endif