From db0397cfaaf488364e3d2e1669dfefae2ee6ea73 Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Mon, 14 Dec 2015 19:57:10 +0000
Subject: [PATCH] shortcut layers, msr networks
---
src/data.c | 513 +++++++++++++++++++++++++++++++++++++++++++++-----------
1 files changed, 413 insertions(+), 100 deletions(-)
diff --git a/src/data.c b/src/data.c
index 012d7cf..8d76276 100644
--- a/src/data.c
+++ b/src/data.c
@@ -1,6 +1,7 @@
#include "data.h"
#include "utils.h"
#include "image.h"
+#include "cuda.h"
#include <stdio.h>
#include <stdlib.h>
@@ -8,22 +9,6 @@
unsigned int data_seed;
-struct load_args{
- char **paths;
- int n;
- int m;
- char **labels;
- int k;
- int h;
- int w;
- int nh;
- int nw;
- int num_boxes;
- int classes;
- int background;
- data *d;
-};
-
list *get_paths(char *filename)
{
char *path;
@@ -49,6 +34,38 @@
return random_paths;
}
+char **find_replace_paths(char **paths, int n, char *find, char *replace)
+{
+ char **replace_paths = calloc(n, sizeof(char*));
+ int i;
+ for(i = 0; i < n; ++i){
+ char *replaced = find_replace(paths[i], find, replace);
+ replace_paths[i] = copy_string(replaced);
+ }
+ return replace_paths;
+}
+
+matrix load_image_paths_gray(char **paths, int n, int w, int h)
+{
+ int i;
+ matrix X;
+ X.rows = n;
+ X.vals = calloc(X.rows, sizeof(float*));
+ X.cols = 0;
+
+ for(i = 0; i < n; ++i){
+ image im = load_image(paths[i], w, h, 3);
+
+ image gray = grayscale_image(im);
+ free_image(im);
+ im = gray;
+
+ X.vals[i] = im.data;
+ X.cols = im.h*im.w*im.c;
+ }
+ return X;
+}
+
matrix load_image_paths(char **paths, int n, int w, int h)
{
int i;
@@ -65,22 +82,16 @@
return X;
}
-typedef struct box{
- int id;
- float x,y,w,h;
- float left, right, top, bottom;
-} box;
-
-box *read_boxes(char *filename, int *n)
+box_label *read_boxes(char *filename, int *n)
{
- box *boxes = calloc(1, sizeof(box));
+ box_label *boxes = calloc(1, sizeof(box_label));
FILE *file = fopen(filename, "r");
if(!file) file_error(filename);
float x, y, h, w;
int id;
int count = 0;
while(fscanf(file, "%d %f %f %f %f", &id, &x, &y, &w, &h) == 5){
- boxes = realloc(boxes, (count+1)*sizeof(box));
+ boxes = realloc(boxes, (count+1)*sizeof(box_label));
boxes[count].id = id;
boxes[count].x = x;
boxes[count].y = y;
@@ -97,24 +108,137 @@
return boxes;
}
-void randomize_boxes(box *b, int n)
+void randomize_boxes(box_label *b, int n)
{
int i;
for(i = 0; i < n; ++i){
- box swap = b[i];
+ box_label swap = b[i];
int index = rand_r(&data_seed)%n;
b[i] = b[index];
b[index] = swap;
}
}
+void correct_boxes(box_label *boxes, int n, float dx, float dy, float sx, float sy, int flip)
+{
+ int i;
+ for(i = 0; i < n; ++i){
+ boxes[i].left = boxes[i].left * sx - dx;
+ boxes[i].right = boxes[i].right * sx - dx;
+ boxes[i].top = boxes[i].top * sy - dy;
+ boxes[i].bottom = boxes[i].bottom* sy - dy;
+
+ if(flip){
+ float swap = boxes[i].left;
+ boxes[i].left = 1. - boxes[i].right;
+ boxes[i].right = 1. - swap;
+ }
+
+ boxes[i].left = constrain(0, 1, boxes[i].left);
+ boxes[i].right = constrain(0, 1, boxes[i].right);
+ boxes[i].top = constrain(0, 1, boxes[i].top);
+ boxes[i].bottom = constrain(0, 1, boxes[i].bottom);
+
+ boxes[i].x = (boxes[i].left+boxes[i].right)/2;
+ boxes[i].y = (boxes[i].top+boxes[i].bottom)/2;
+ boxes[i].w = (boxes[i].right - boxes[i].left);
+ boxes[i].h = (boxes[i].bottom - boxes[i].top);
+
+ boxes[i].w = constrain(0, 1, boxes[i].w);
+ boxes[i].h = constrain(0, 1, boxes[i].h);
+ }
+}
+
+void fill_truth_swag(char *path, float *truth, int classes, int flip, float dx, float dy, float sx, float sy)
+{
+ char *labelpath = find_replace(path, "images", "labels");
+ labelpath = find_replace(labelpath, "JPEGImages", "labels");
+
+ labelpath = find_replace(labelpath, ".jpg", ".txt");
+ labelpath = find_replace(labelpath, ".JPG", ".txt");
+ labelpath = find_replace(labelpath, ".JPEG", ".txt");
+ int count = 0;
+ box_label *boxes = read_boxes(labelpath, &count);
+ randomize_boxes(boxes, count);
+ correct_boxes(boxes, count, dx, dy, sx, sy, flip);
+ float x,y,w,h;
+ int id;
+ int i;
+
+ for (i = 0; i < count && i < 30; ++i) {
+ x = boxes[i].x;
+ y = boxes[i].y;
+ w = boxes[i].w;
+ h = boxes[i].h;
+ id = boxes[i].id;
+
+ if (w < .0 || h < .0) continue;
+
+ int index = (4+classes) * i;
+
+ truth[index++] = x;
+ truth[index++] = y;
+ truth[index++] = w;
+ truth[index++] = h;
+
+ if (id < classes) truth[index+id] = 1;
+ }
+ free(boxes);
+}
+
+void fill_truth_region(char *path, float *truth, int classes, int num_boxes, int flip, float dx, float dy, float sx, float sy)
+{
+ char *labelpath = find_replace(path, "images", "labels");
+ labelpath = find_replace(labelpath, "JPEGImages", "labels");
+
+ labelpath = find_replace(labelpath, ".jpg", ".txt");
+ labelpath = find_replace(labelpath, ".JPG", ".txt");
+ labelpath = find_replace(labelpath, ".JPEG", ".txt");
+ int count = 0;
+ box_label *boxes = read_boxes(labelpath, &count);
+ randomize_boxes(boxes, count);
+ correct_boxes(boxes, count, dx, dy, sx, sy, flip);
+ float x,y,w,h;
+ int id;
+ int i;
+
+ for (i = 0; i < count; ++i) {
+ x = boxes[i].x;
+ y = boxes[i].y;
+ w = boxes[i].w;
+ h = boxes[i].h;
+ id = boxes[i].id;
+
+ if (w < .01 || h < .01) continue;
+
+ int col = (int)(x*num_boxes);
+ int row = (int)(y*num_boxes);
+
+ x = x*num_boxes - col;
+ y = y*num_boxes - row;
+
+ int index = (col+row*num_boxes)*(5+classes);
+ if (truth[index]) continue;
+ truth[index++] = 1;
+
+ if (id < classes) truth[index+id] = 1;
+ index += classes;
+
+ truth[index++] = x;
+ truth[index++] = y;
+ truth[index++] = w;
+ truth[index++] = h;
+ }
+ free(boxes);
+}
+
void fill_truth_detection(char *path, float *truth, int classes, int num_boxes, int flip, int background, float dx, float dy, float sx, float sy)
{
- char *labelpath = find_replace(path, "detection_images", "labels");
+ char *labelpath = find_replace(path, "JPEGImages", "labels");
labelpath = find_replace(labelpath, ".jpg", ".txt");
labelpath = find_replace(labelpath, ".JPEG", ".txt");
int count = 0;
- box *boxes = read_boxes(labelpath, &count);
+ box_label *boxes = read_boxes(labelpath, &count);
randomize_boxes(boxes, count);
float x,y,w,h;
float left, top, right, bot;
@@ -148,36 +272,38 @@
w = (right - left);
h = (bot - top);
- if (x <= 0 || x >= 1 || y <= 0 || y >= 1) continue;
+ if (x <= 0 || x >= 1 || y <= 0 || y >= 1) continue;
- int i = (int)(x*num_boxes);
- int j = (int)(y*num_boxes);
+ int col = (int)(x*num_boxes);
+ int row = (int)(y*num_boxes);
- x = x*num_boxes - i;
- y = y*num_boxes - j;
+ x = x*num_boxes - col;
+ y = y*num_boxes - row;
/*
- float maxwidth = distance_from_edge(i, num_boxes);
- float maxheight = distance_from_edge(j, num_boxes);
- w = w/maxwidth;
- h = h/maxheight;
- */
+ float maxwidth = distance_from_edge(i, num_boxes);
+ float maxheight = distance_from_edge(j, num_boxes);
+ w = w/maxwidth;
+ h = h/maxheight;
+ */
w = constrain(0, 1, w);
h = constrain(0, 1, h);
- if (w == 0 || h == 0) continue;
- w = sqrt(w);
- h = sqrt(h);
+ if (w < .01 || h < .01) continue;
+ if(1){
+ w = pow(w, 1./2.);
+ h = pow(h, 1./2.);
+ }
- int index = (i+j*num_boxes)*(4+classes+background);
+ int index = (col+row*num_boxes)*(4+classes+background);
if(truth[index+classes+background+2]) continue;
if(background) truth[index++] = 0;
truth[index+id] = 1;
index += classes;
- truth[index++] = y;
truth[index++] = x;
- truth[index++] = h;
+ truth[index++] = y;
truth[index++] = w;
+ truth[index++] = h;
}
free(boxes);
}
@@ -247,7 +373,7 @@
++count;
}
}
- if(count != 1) printf("%d, %s\n", count, path);
+ if(count != 1) printf("Too many or too few labels: %d, %s\n", count, path);
}
matrix load_labels_paths(char **paths, int n, char **labels, int k)
@@ -279,7 +405,174 @@
}
}
-data load_data_detection_jitter_random(int n, char **paths, int m, int classes, int w, int h, int num_boxes, int background)
+data load_data_region(int n, char **paths, int m, int w, int h, int size, int classes, float jitter)
+{
+ char **random_paths = get_random_paths(paths, n, m);
+ int i;
+ data d;
+ d.shallow = 0;
+
+ d.X.rows = n;
+ d.X.vals = calloc(d.X.rows, sizeof(float*));
+ d.X.cols = h*w*3;
+
+ int k = size*size*(5+classes);
+ d.y = make_matrix(n, k);
+ for(i = 0; i < n; ++i){
+ image orig = load_image_color(random_paths[i], 0, 0);
+
+ int oh = orig.h;
+ int ow = orig.w;
+
+ int dw = (ow*jitter);
+ int dh = (oh*jitter);
+
+ int pleft = (rand_uniform() * 2*dw - dw);
+ int pright = (rand_uniform() * 2*dw - dw);
+ int ptop = (rand_uniform() * 2*dh - dh);
+ int pbot = (rand_uniform() * 2*dh - dh);
+
+ int swidth = ow - pleft - pright;
+ int sheight = oh - ptop - pbot;
+
+ float sx = (float)swidth / ow;
+ float sy = (float)sheight / oh;
+
+ int flip = rand_r(&data_seed)%2;
+ image cropped = crop_image(orig, pleft, ptop, swidth, sheight);
+
+ float dx = ((float)pleft/ow)/sx;
+ float dy = ((float)ptop /oh)/sy;
+
+ image sized = resize_image(cropped, w, h);
+ if(flip) flip_image(sized);
+ d.X.vals[i] = sized.data;
+
+ fill_truth_region(random_paths[i], d.y.vals[i], classes, size, flip, dx, dy, 1./sx, 1./sy);
+
+ free_image(orig);
+ free_image(cropped);
+ }
+ free(random_paths);
+ return d;
+}
+
+data load_data_compare(int n, char **paths, int m, int classes, int w, int h)
+{
+ if(m) paths = get_random_paths(paths, 2*n, m);
+ int i,j;
+ data d;
+ d.shallow = 0;
+
+ d.X.rows = n;
+ d.X.vals = calloc(d.X.rows, sizeof(float*));
+ d.X.cols = h*w*6;
+
+ int k = 2*(classes);
+ d.y = make_matrix(n, k);
+ for(i = 0; i < n; ++i){
+ image im1 = load_image_color(paths[i*2], w, h);
+ image im2 = load_image_color(paths[i*2+1], w, h);
+
+ d.X.vals[i] = calloc(d.X.cols, sizeof(float));
+ memcpy(d.X.vals[i], im1.data, h*w*3*sizeof(float));
+ memcpy(d.X.vals[i] + h*w*3, im2.data, h*w*3*sizeof(float));
+
+ int id;
+ float iou;
+
+ char *imlabel1 = find_replace(paths[i*2], "imgs", "labels");
+ imlabel1 = find_replace(imlabel1, "jpg", "txt");
+ FILE *fp1 = fopen(imlabel1, "r");
+
+ while(fscanf(fp1, "%d %f", &id, &iou) == 2){
+ if (d.y.vals[i][2*id] < iou) d.y.vals[i][2*id] = iou;
+ }
+
+ char *imlabel2 = find_replace(paths[i*2+1], "imgs", "labels");
+ imlabel2 = find_replace(imlabel2, "jpg", "txt");
+ FILE *fp2 = fopen(imlabel2, "r");
+
+ while(fscanf(fp2, "%d %f", &id, &iou) == 2){
+ if (d.y.vals[i][2*id + 1] < iou) d.y.vals[i][2*id + 1] = iou;
+ }
+
+ for (j = 0; j < classes; ++j){
+ if (d.y.vals[i][2*j] > .5 && d.y.vals[i][2*j+1] < .5){
+ d.y.vals[i][2*j] = 1;
+ d.y.vals[i][2*j+1] = 0;
+ } else if (d.y.vals[i][2*j] < .5 && d.y.vals[i][2*j+1] > .5){
+ d.y.vals[i][2*j] = 0;
+ d.y.vals[i][2*j+1] = 1;
+ } else {
+ d.y.vals[i][2*j] = SECRET_NUM;
+ d.y.vals[i][2*j+1] = SECRET_NUM;
+ }
+ }
+ fclose(fp1);
+ fclose(fp2);
+
+ free_image(im1);
+ free_image(im2);
+ }
+ if(m) free(paths);
+ return d;
+}
+
+data load_data_swag(char **paths, int n, int classes, float jitter)
+{
+ int index = rand_r(&data_seed)%n;
+ char *random_path = paths[index];
+
+ image orig = load_image_color(random_path, 0, 0);
+ int h = orig.h;
+ int w = orig.w;
+
+ data d;
+ d.shallow = 0;
+ d.w = w;
+ d.h = h;
+
+ d.X.rows = 1;
+ d.X.vals = calloc(d.X.rows, sizeof(float*));
+ d.X.cols = h*w*3;
+
+ int k = (4+classes)*30;
+ d.y = make_matrix(1, k);
+
+ int dw = w*jitter;
+ int dh = h*jitter;
+
+ int pleft = (rand_uniform() * 2*dw - dw);
+ int pright = (rand_uniform() * 2*dw - dw);
+ int ptop = (rand_uniform() * 2*dh - dh);
+ int pbot = (rand_uniform() * 2*dh - dh);
+
+ int swidth = w - pleft - pright;
+ int sheight = h - ptop - pbot;
+
+ float sx = (float)swidth / w;
+ float sy = (float)sheight / h;
+
+ int flip = rand_r(&data_seed)%2;
+ image cropped = crop_image(orig, pleft, ptop, swidth, sheight);
+
+ float dx = ((float)pleft/w)/sx;
+ float dy = ((float)ptop /h)/sy;
+
+ image sized = resize_image(cropped, w, h);
+ if(flip) flip_image(sized);
+ d.X.vals[0] = sized.data;
+
+ fill_truth_swag(random_path, d.y.vals[0], classes, flip, dx, dy, 1./sx, 1./sy);
+
+ free_image(orig);
+ free_image(cropped);
+
+ return d;
+}
+
+data load_data_detection(int n, char **paths, int m, int classes, int w, int h, int num_boxes, int background)
{
char **random_paths = get_random_paths(paths, n, m);
int i;
@@ -294,11 +587,6 @@
d.y = make_matrix(n, k);
for(i = 0; i < n; ++i){
image orig = load_image_color(random_paths[i], 0, 0);
- float exposure = rand_uniform()+1;
- if(rand_uniform() > .5) exposure = 1/exposure;
-
- float saturation = rand_uniform()+1;
- if(rand_uniform() > .5) saturation = 1/saturation;
int oh = orig.h;
int ow = orig.w;
@@ -316,61 +604,87 @@
float sx = (float)swidth / ow;
float sy = (float)sheight / oh;
-
+
/*
- float angle = rand_uniform()*.1 - .05;
- image rot = rotate_image(orig, angle);
- free_image(orig);
- orig = rot;
- */
+ float angle = rand_uniform()*.1 - .05;
+ image rot = rotate_image(orig, angle);
+ free_image(orig);
+ orig = rot;
+ */
int flip = rand_r(&data_seed)%2;
image cropped = crop_image(orig, pleft, ptop, swidth, sheight);
+
float dx = ((float)pleft/ow)/sx;
float dy = ((float)ptop /oh)/sy;
- free_image(orig);
image sized = resize_image(cropped, w, h);
- free_image(cropped);
if(flip) flip_image(sized);
d.X.vals[i] = sized.data;
fill_truth_detection(random_paths[i], d.y.vals[i], classes, num_boxes, flip, background, dx, dy, 1./sx, 1./sy);
+
+ free_image(orig);
+ free_image(cropped);
}
free(random_paths);
return d;
}
-void *load_detection_thread(void *ptr)
+void *load_thread(void *ptr)
{
- printf("Loading data: %d\n", rand_r(&data_seed));
- struct load_args a = *(struct load_args*)ptr;
- *a.d = load_data_detection_jitter_random(a.n, a.paths, a.m, a.classes, a.w, a.h, a.num_boxes, a.background);
+
+#ifdef GPU
+ cudaError_t status = cudaSetDevice(gpu_index);
+ check_error(status);
+#endif
+
+ //printf("Loading data: %d\n", rand_r(&data_seed));
+ load_args a = *(struct load_args*)ptr;
+ if (a.type == CLASSIFICATION_DATA){
+ *a.d = load_data(a.paths, a.n, a.m, a.labels, a.classes, a.w, a.h);
+ } else if (a.type == DETECTION_DATA){
+ *a.d = load_data_detection(a.n, a.paths, a.m, a.classes, a.w, a.h, a.num_boxes, a.background);
+ } else if (a.type == WRITING_DATA){
+ *a.d = load_data_writing(a.paths, a.n, a.m, a.w, a.h, a.out_w, a.out_h);
+ } else if (a.type == REGION_DATA){
+ *a.d = load_data_region(a.n, a.paths, a.m, a.w, a.h, a.num_boxes, a.classes, a.jitter);
+ } else if (a.type == SWAG_DATA){
+ *a.d = load_data_swag(a.paths, a.n, a.classes, a.jitter);
+ } else if (a.type == COMPARE_DATA){
+ *a.d = load_data_compare(a.n, a.paths, a.m, a.classes, a.w, a.h);
+ } else if (a.type == IMAGE_DATA){
+ *(a.im) = load_image_color(a.path, 0, 0);
+ *(a.resized) = resize_image(*(a.im), a.w, a.h);
+ }
free(ptr);
return 0;
}
-pthread_t load_data_detection_thread(int n, char **paths, int m, int classes, int w, int h, int nh, int nw, int background, data *d)
+pthread_t load_data_in_thread(load_args args)
{
pthread_t thread;
- struct load_args *args = calloc(1, sizeof(struct load_args));
- args->n = n;
- args->paths = paths;
- args->m = m;
- args->h = h;
- args->w = w;
- args->nh = nh;
- args->nw = nw;
- args->num_boxes = nw;
- args->classes = classes;
- args->background = background;
- args->d = d;
- if(pthread_create(&thread, 0, load_detection_thread, args)) {
- error("Thread creation failed");
- }
+ struct load_args *ptr = calloc(1, sizeof(struct load_args));
+ *ptr = args;
+ if(pthread_create(&thread, 0, load_thread, ptr)) error("Thread creation failed");
return thread;
}
+data load_data_writing(char **paths, int n, int m, int w, int h, int out_w, int out_h)
+{
+ if(m) paths = get_random_paths(paths, n, m);
+ char **replace_paths = find_replace_paths(paths, n, ".png", "-label.png");
+ data d;
+ d.shallow = 0;
+ d.X = load_image_paths(paths, n, w, h);
+ d.y = load_image_paths_gray(replace_paths, n, out_w, out_h);
+ if(m) free(paths);
+ int i;
+ for(i = 0; i < n; ++i) free(replace_paths[i]);
+ free(replace_paths);
+ return d;
+}
+
data load_data(char **paths, int n, int m, char **labels, int k, int w, int h)
{
if(m) paths = get_random_paths(paths, n, m);
@@ -382,30 +696,29 @@
return d;
}
-void *load_in_thread(void *ptr)
+matrix concat_matrix(matrix m1, matrix m2)
{
- struct load_args a = *(struct load_args*)ptr;
- *a.d = load_data(a.paths, a.n, a.m, a.labels, a.k, a.w, a.h);
- free(ptr);
- return 0;
+ int i, count = 0;
+ matrix m;
+ m.cols = m1.cols;
+ m.rows = m1.rows+m2.rows;
+ m.vals = calloc(m1.rows + m2.rows, sizeof(float*));
+ for(i = 0; i < m1.rows; ++i){
+ m.vals[count++] = m1.vals[i];
+ }
+ for(i = 0; i < m2.rows; ++i){
+ m.vals[count++] = m2.vals[i];
+ }
+ return m;
}
-pthread_t load_data_thread(char **paths, int n, int m, char **labels, int k, int w, int h, data *d)
+data concat_data(data d1, data d2)
{
- pthread_t thread;
- struct load_args *args = calloc(1, sizeof(struct load_args));
- args->n = n;
- args->paths = paths;
- args->m = m;
- args->labels = labels;
- args->k = k;
- args->h = h;
- args->w = w;
- args->d = d;
- if(pthread_create(&thread, 0, load_in_thread, args)) {
- error("Thread creation failed");
- }
- return thread;
+ data d;
+ d.shallow = 1;
+ d.X = concat_matrix(d1.X, d2.X);
+ d.y = concat_matrix(d1.y, d2.y);
+ return d;
}
data load_categorical_data_csv(char *filename, int target, int k)
--
Gitblit v1.10.0