From 1b5afb45838e603fa6780762eb8cc59246dc2d81 Mon Sep 17 00:00:00 2001
From: IlyaOvodov <b@ovdv.ru>
Date: Tue, 08 May 2018 11:09:35 +0000
Subject: [PATCH] Output improvements for detector results: When printing detector results, output was done in random order, obfuscating results for interpreting. Now: 1. Text output includes coordinates of rects in (left,right,top,bottom in pixels) along with label and score 2. Text output is sorted by rect lefts to simplify finding appropriate rects on image 3. If several class probs are > thresh for some detection, the most probable is written first and coordinates for others are not repeated 4. Rects are imprinted in image in order by their best class prob, so most probable rects are always on top and not overlayed by less probable ones 5. Most probable label for rect is always written first Also: 6. Message about low GPU memory include required amount
---
src/data.c | 1095 +++++++++++++++++++++++++++++++++++++++++++--------------
1 files changed, 827 insertions(+), 268 deletions(-)
diff --git a/src/data.c b/src/data.c
index 0184984..3b014b4 100644
--- a/src/data.c
+++ b/src/data.c
@@ -1,28 +1,13 @@
#include "data.h"
#include "utils.h"
#include "image.h"
+#include "cuda.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
-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;
-};
+pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
list *get_paths(char *filename)
{
@@ -37,18 +22,72 @@
return lines;
}
+/*
+char **get_random_paths_indexes(char **paths, int n, int m, int *indexes)
+{
+ char **random_paths = calloc(n, sizeof(char*));
+ int i;
+ pthread_mutex_lock(&mutex);
+ for(i = 0; i < n; ++i){
+ int index = random_gen()%m;
+ indexes[i] = index;
+ random_paths[i] = paths[index];
+ if(i == 0) printf("%s\n", paths[index]);
+ }
+ pthread_mutex_unlock(&mutex);
+ return random_paths;
+}
+*/
+
char **get_random_paths(char **paths, int n, int m)
{
char **random_paths = calloc(n, sizeof(char*));
int i;
- for(i = 0; i < n; ++i){
- int index = rand_r(&data_seed)%m;
+ pthread_mutex_lock(&mutex);
+ //printf("n = %d \n", n);
+ for(i = 0; i < n; ++i){
+ int index = random_gen() % m;
random_paths[i] = paths[index];
- if(i == 0) printf("%s\n", paths[index]);
+ //if(i == 0) printf("%s\n", paths[index]);
+ //printf("grp: %s\n", paths[index]);
}
+ pthread_mutex_unlock(&mutex);
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[4096];
+ find_replace(paths[i], find, replace, replaced);
+ 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;
@@ -65,11 +104,34 @@
return X;
}
-typedef struct{
- int id;
- float x,y,w,h;
- float left, right, top, bottom;
-} box_label;
+matrix load_image_augment_paths(char **paths, int n, int use_flip, int min, int max, int size, float angle, float aspect, float hue, float saturation, float exposure)
+{
+ 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_color(paths[i], 0, 0);
+ image crop = random_augment_image(im, angle, aspect, min, max, size);
+ int flip = use_flip ? random_gen() % 2 : 0;
+ if (flip)
+ flip_image(crop);
+ random_distort_image(crop, hue, saturation, exposure);
+
+ /*
+ show_image(im, "orig");
+ show_image(crop, "crop");
+ cvWaitKey(0);
+ */
+ free_image(im);
+ X.vals[i] = crop.data;
+ X.cols = crop.h*crop.w*crop.c;
+ }
+ return X;
+}
+
box_label *read_boxes(char *filename, int *n)
{
@@ -102,140 +164,185 @@
int i;
for(i = 0; i < n; ++i){
box_label swap = b[i];
- int index = rand_r(&data_seed)%n;
+ int index = random_gen()%n;
b[i] = b[index];
b[index] = swap;
}
}
-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)
+void correct_boxes(box_label *boxes, int n, float dx, float dy, float sx, float sy, int flip)
{
- char *labelpath = find_replace(path, "detection_images", "labels");
- labelpath = find_replace(labelpath, ".jpg", ".txt");
- labelpath = find_replace(labelpath, ".JPEG", ".txt");
+ int i;
+ for(i = 0; i < n; ++i){
+ if(boxes[i].x == 0 && boxes[i].y == 0) {
+ boxes[i].x = 999999;
+ boxes[i].y = 999999;
+ boxes[i].w = 999999;
+ boxes[i].h = 999999;
+ continue;
+ }
+ 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[4096];
+ 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", labelpath);
+
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;
- float left, top, right, bot;
int id;
int i;
- if(background){
- for(i = 0; i < num_boxes*num_boxes*(4+classes+background); i += 4+classes+background){
- truth[i] = 1;
- }
- }
- for(i = 0; i < count; ++i){
- left = boxes[i].left * sx - dx;
- right = boxes[i].right * sx - dx;
- top = boxes[i].top * sy - dy;
- bot = boxes[i].bottom* sy - dy;
+
+ 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(flip){
- float swap = left;
- left = 1. - right;
- right = 1. - swap;
- }
+ if (w < .0 || h < .0) continue;
- left = constrain(0, 1, left);
- right = constrain(0, 1, right);
- top = constrain(0, 1, top);
- bot = constrain(0, 1, bot);
+ int index = (4+classes) * i;
- x = (left+right)/2;
- y = (top+bot)/2;
- w = (right - left);
- h = (bot - top);
-
- if (x <= 0 || x >= 1 || y <= 0 || y >= 1) continue;
-
- int i = (int)(x*num_boxes);
- int j = (int)(y*num_boxes);
-
- x = x*num_boxes - i;
- y = y*num_boxes - j;
-
- /*
- 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 < .01 || h < .01) continue;
- if(1){
- //w = sqrt(w);
- //h = sqrt(h);
- w = pow(w, 1./2.);
- h = pow(h, 1./2.);
- }
-
- int index = (i+j*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++] = x;
truth[index++] = y;
truth[index++] = w;
truth[index++] = h;
+
+ if (id < classes) truth[index+id] = 1;
}
free(boxes);
}
-void fill_truth_localization(char *path, float *truth, int classes, int flip, float dx, float dy, float sx, float sy)
+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, "objects", "object_labels");
- labelpath = find_replace(labelpath, ".jpg", ".txt");
- labelpath = find_replace(labelpath, ".JPEG", ".txt");
- int count;
+ char labelpath[4096];
+ find_replace(path, "images", "labels", labelpath);
+ find_replace(labelpath, "JPEGImages", "labels", labelpath);
+
+ find_replace(labelpath, ".jpg", ".txt", labelpath);
+ find_replace(labelpath, ".png", ".txt", labelpath);
+ find_replace(labelpath, ".JPG", ".txt", labelpath);
+ find_replace(labelpath, ".JPEG", ".txt", labelpath);
+ int count = 0;
box_label *boxes = read_boxes(labelpath, &count);
- box_label box = boxes[0];
- free(boxes);
+ randomize_boxes(boxes, count);
+ correct_boxes(boxes, count, dx, dy, sx, sy, flip);
float x,y,w,h;
- float left, top, right, bot;
int id;
int i;
- for(i = 0; i < count; ++i){
- left = box.left * sx - dx;
- right = box.right * sx - dx;
- top = box.top * sy - dy;
- bot = box.bottom* sy - dy;
- id = box.id;
- if(flip){
- float swap = left;
- left = 1. - right;
- right = 1. - swap;
- }
+ 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;
- left = constrain(0, 1, left);
- right = constrain(0, 1, right);
- top = constrain(0, 1, top);
- bot = constrain(0, 1, bot);
+ if (w < .001 || h < .001) continue;
- x = (left+right)/2;
- y = (top+bot)/2;
- w = (right - left);
- h = (bot - top);
+ int col = (int)(x*num_boxes);
+ int row = (int)(y*num_boxes);
- if (x <= 0 || x >= 1 || y <= 0 || y >= 1) continue;
+ x = x*num_boxes - col;
+ y = y*num_boxes - row;
- w = constrain(0, 1, w);
- h = constrain(0, 1, h);
- if (w == 0 || h == 0) continue;
+ int index = (col+row*num_boxes)*(5+classes);
+ if (truth[index]) continue;
+ truth[index++] = 1;
- int index = id*4;
+ 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, int num_boxes, float *truth, int classes, int flip, float dx, float dy, float sx, float sy,
+ int small_object, int net_w, int net_h)
+{
+ char labelpath[4096];
+ find_replace(path, "images", "labels", labelpath);
+ find_replace(labelpath, "JPEGImages", "labels", labelpath);
+
+ find_replace(labelpath, "raw", "labels", labelpath);
+ find_replace(labelpath, ".jpg", ".txt", labelpath);
+ find_replace(labelpath, ".png", ".txt", labelpath);
+ find_replace(labelpath, ".bmp", ".txt", labelpath);
+ find_replace(labelpath, ".JPG", ".txt", labelpath);
+ find_replace(labelpath, ".JPEG", ".txt", labelpath);
+ int count = 0;
+ int i;
+ box_label *boxes = read_boxes(labelpath, &count);
+ float lowest_w = 1.F / net_w;
+ float lowest_h = 1.F / net_h;
+ if (small_object == 1) {
+ for (i = 0; i < count; ++i) {
+ if (boxes[i].w < lowest_w) boxes[i].w = lowest_w;
+ if (boxes[i].h < lowest_h) boxes[i].h = lowest_h;
+ }
+ }
+ randomize_boxes(boxes, count);
+ correct_boxes(boxes, count, dx, dy, sx, sy, flip);
+ if(count > num_boxes) count = num_boxes;
+ float x,y,w,h;
+ int id;
+
+ 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;
+
+ // not detect small objects
+ //if ((w < 0.001F || h < 0.001F)) continue;
+ // if truth (box for object) is smaller than 1x1 pix
+ if ((w < lowest_w || h < lowest_h)) continue;
+
+ truth[i*5+0] = x;
+ truth[i*5+1] = y;
+ truth[i*5+2] = w;
+ truth[i*5+3] = h;
+ truth[i*5+4] = id;
+ }
+ free(boxes);
+}
#define NUMCHARS 37
@@ -267,7 +374,7 @@
data load_data_captcha(char **paths, int n, int m, int k, int w, int h)
{
if(m) paths = get_random_paths(paths, n, m);
- data d;
+ data d = {0};
d.shallow = 0;
d.X = load_image_paths(paths, n, w, h);
d.y = make_matrix(n, k*NUMCHARS);
@@ -282,7 +389,7 @@
data load_data_captcha_encode(char **paths, int n, int m, int w, int h)
{
if(m) paths = get_random_paths(paths, n, m);
- data d;
+ data d = {0};
d.shallow = 0;
d.X = load_image_paths(paths, n, w, h);
d.X.cols = 17100;
@@ -302,19 +409,82 @@
++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)
+void fill_hierarchy(float *truth, int k, tree *hierarchy)
+{
+ int j;
+ for(j = 0; j < k; ++j){
+ if(truth[j]){
+ int parent = hierarchy->parent[j];
+ while(parent >= 0){
+ truth[parent] = 1;
+ parent = hierarchy->parent[parent];
+ }
+ }
+ }
+ int i;
+ int count = 0;
+ for(j = 0; j < hierarchy->groups; ++j){
+ //printf("%d\n", count);
+ int mask = 1;
+ for(i = 0; i < hierarchy->group_size[j]; ++i){
+ if(truth[count + i]){
+ mask = 0;
+ break;
+ }
+ }
+ if (mask) {
+ for(i = 0; i < hierarchy->group_size[j]; ++i){
+ truth[count + i] = SECRET_NUM;
+ }
+ }
+ count += hierarchy->group_size[j];
+ }
+}
+
+matrix load_labels_paths(char **paths, int n, char **labels, int k, tree *hierarchy)
{
matrix y = make_matrix(n, k);
int i;
for(i = 0; i < n && labels; ++i){
fill_truth(paths[i], labels, k, y.vals[i]);
+ if(hierarchy){
+ fill_hierarchy(y.vals[i], k, hierarchy);
+ }
}
return y;
}
+matrix load_tags_paths(char **paths, int n, int k)
+{
+ matrix y = make_matrix(n, k);
+ int i;
+ int count = 0;
+ for(i = 0; i < n; ++i){
+ char label[4096];
+ find_replace(paths[i], "imgs", "labels", label);
+ find_replace(label, "_iconl.jpeg", ".txt", label);
+ FILE *file = fopen(label, "r");
+ if(!file){
+ find_replace(label, "labels", "labels2", label);
+ file = fopen(label, "r");
+ if(!file) continue;
+ }
+ ++count;
+ int tag;
+ while(fscanf(file, "%d", &tag) == 1){
+ if(tag < k){
+ y.vals[i][tag] = 1;
+ }
+ }
+ fclose(file);
+ }
+ printf("%d/%d\n", count, n);
+ return y;
+}
+
char **get_labels(char *filename)
{
list *plist = get_paths(filename);
@@ -334,18 +504,19 @@
}
}
-data load_data_localization(int n, char **paths, int m, int classes, int w, int h)
+data load_data_region(int n, char **paths, int m, int w, int h, int size, int classes, float jitter, float hue, float saturation, float exposure)
{
char **random_paths = get_random_paths(paths, n, m);
int i;
- data d;
+ data d = {0};
d.shallow = 0;
d.X.rows = n;
d.X.vals = calloc(d.X.rows, sizeof(float*));
d.X.cols = h*w*3;
- int k = (4*classes);
+
+ 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);
@@ -353,13 +524,13 @@
int oh = orig.h;
int ow = orig.w;
- int dw = 32;
- int dh = 32;
+ int dw = (ow*jitter);
+ int dh = (oh*jitter);
- int pleft = (rand_uniform() * dw);
- int pright = (rand_uniform() * dw);
- int ptop = (rand_uniform() * dh);
- int pbot = (rand_uniform() * dh);
+ int pleft = rand_uniform(-dw, dw);
+ int pright = rand_uniform(-dw, dw);
+ int ptop = rand_uniform(-dh, dh);
+ int pbot = rand_uniform(-dh, dh);
int swidth = ow - pleft - pright;
int sheight = oh - ptop - pbot;
@@ -367,173 +538,454 @@
float sx = (float)swidth / ow;
float sy = (float)sheight / oh;
- int flip = rand_r(&data_seed)%2;
+ int flip = random_gen()%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);
+ random_distort_image(sized, hue, saturation, exposure);
d.X.vals[i] = sized.data;
- fill_truth_localization(random_paths[i], d.y.vals[i], classes, flip, dx, dy, 1./sx, 1./sy);
+ 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_detection_jitter_random(int n, char **paths, int m, int classes, int w, int h, int num_boxes, int background)
+data load_data_compare(int n, char **paths, int m, int classes, int w, int h)
{
- char **random_paths = get_random_paths(paths, n, m);
- int i;
- data d;
+ if(m) paths = get_random_paths(paths, 2*n, m);
+ int i,j;
+ data d = {0};
d.shallow = 0;
d.X.rows = n;
d.X.vals = calloc(d.X.rows, sizeof(float*));
- d.X.cols = h*w*3;
+ d.X.cols = h*w*6;
- int k = num_boxes*num_boxes*(4+classes+background);
+ int k = 2*(classes);
d.y = make_matrix(n, k);
for(i = 0; i < n; ++i){
- image orig = load_image_color(random_paths[i], 0, 0);
+ image im1 = load_image_color(paths[i*2], w, h);
+ image im2 = load_image_color(paths[i*2+1], w, h);
- int oh = orig.h;
- int ow = orig.w;
+ 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 dw = ow/10;
- int dh = oh/10;
+ int id;
+ float iou;
- 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);
+ char imlabel1[4096];
+ char imlabel2[4096];
+ find_replace(paths[i*2], "imgs", "labels", imlabel1);
+ find_replace(imlabel1, "jpg", "txt", imlabel1);
+ FILE *fp1 = fopen(imlabel1, "r");
- int swidth = ow - pleft - pright;
- int sheight = oh - ptop - pbot;
+ while(fscanf(fp1, "%d %f", &id, &iou) == 2){
+ if (d.y.vals[i][2*id] < iou) d.y.vals[i][2*id] = iou;
+ }
- float sx = (float)swidth / ow;
- float sy = (float)sheight / oh;
+ find_replace(paths[i*2+1], "imgs", "labels", imlabel2);
+ find_replace(imlabel2, "jpg", "txt", imlabel2);
+ FILE *fp2 = fopen(imlabel2, "r");
- /*
- float angle = rand_uniform()*.1 - .05;
- image rot = rotate_image(orig, angle);
- free_image(orig);
- orig = rot;
- */
+ 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;
+ }
- 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;
+ 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(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(im1);
+ free_image(im2);
}
- free(random_paths);
- return d;
-}
-
-void *load_localization_thread(void *ptr)
-{
- printf("Loading data: %d\n", rand_r(&data_seed));
- struct load_args a = *(struct load_args*)ptr;
- *a.d = load_data_localization(a.n, a.paths, a.m, a.classes, a.w, a.h);
- free(ptr);
- return 0;
-}
-
-pthread_t load_data_localization_thread(int n, char **paths, int m, int classes, int w, int h, data *d)
-{
- pthread_t thread;
- struct load_args *args = calloc(1, sizeof(struct load_args));
- args->n = n;
- args->paths = paths;
- args->m = m;
- args->w = w;
- args->h = h;
- args->classes = classes;
- args->d = d;
- if(pthread_create(&thread, 0, load_localization_thread, args)) {
- error("Thread creation failed");
- }
- return thread;
-}
-
-void *load_detection_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);
- 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 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");
- }
- return thread;
-}
-
-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);
- data d;
- d.shallow = 0;
- d.X = load_image_paths(paths, n, w, h);
- d.y = load_labels_paths(paths, n, labels, k);
if(m) free(paths);
return d;
}
-void *load_in_thread(void *ptr)
+data load_data_swag(char **paths, int n, int classes, float jitter)
{
- 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);
+ int index = random_gen()%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 = {0};
+ 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(-dw, dw);
+ int pright = rand_uniform(-dw, dw);
+ int ptop = rand_uniform(-dh, dh);
+ int pbot = rand_uniform(-dh, dh);
+
+ int swidth = w - pleft - pright;
+ int sheight = h - ptop - pbot;
+
+ float sx = (float)swidth / w;
+ float sy = (float)sheight / h;
+
+ int flip = random_gen()%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;
+}
+
+#ifdef OPENCV
+#include "opencv2/highgui/highgui_c.h"
+#include "opencv2/imgproc/imgproc_c.h"
+#include "opencv2/core/version.hpp"
+#ifndef CV_VERSION_EPOCH
+#include "opencv2/videoio/videoio_c.h"
+#include "opencv2/imgcodecs/imgcodecs_c.h"
+#endif
+
+#include "http_stream.h"
+
+data load_data_detection(int n, char **paths, int m, int w, int h, int boxes, int classes, int use_flip, float jitter, float hue, float saturation, float exposure, int small_object)
+{
+ char **random_paths = get_random_paths(paths, n, m);
+ int i;
+ data d = {0};
+ d.shallow = 0;
+
+ d.X.rows = n;
+ d.X.vals = calloc(d.X.rows, sizeof(float*));
+ d.X.cols = h*w*3;
+
+ d.y = make_matrix(n, 5*boxes);
+ for(i = 0; i < n; ++i){
+ const char *filename = random_paths[i];
+
+ int flag = 1;
+ IplImage *src;
+ if ((src = cvLoadImage(filename, flag)) == 0)
+ {
+ fprintf(stderr, "Cannot load image \"%s\"\n", filename);
+ char buff[256];
+ sprintf(buff, "echo %s >> bad.list", filename);
+ system(buff);
+ continue;
+ //exit(0);
+ }
+
+ int oh = src->height;
+ int ow = src->width;
+
+ int dw = (ow*jitter);
+ int dh = (oh*jitter);
+
+ int pleft = rand_uniform_strong(-dw, dw);
+ int pright = rand_uniform_strong(-dw, dw);
+ int ptop = rand_uniform_strong(-dh, dh);
+ int pbot = rand_uniform_strong(-dh, dh);
+
+ int swidth = ow - pleft - pright;
+ int sheight = oh - ptop - pbot;
+
+ float sx = (float)swidth / ow;
+ float sy = (float)sheight / oh;
+
+ int flip = use_flip ? random_gen()%2 : 0;
+
+ float dx = ((float)pleft/ow)/sx;
+ float dy = ((float)ptop /oh)/sy;
+
+ float dhue = rand_uniform_strong(-hue, hue);
+ float dsat = rand_scale(saturation);
+ float dexp = rand_scale(exposure);
+
+ image ai = image_data_augmentation(src, w, h, pleft, ptop, swidth, sheight, flip, jitter, dhue, dsat, dexp);
+ d.X.vals[i] = ai.data;
+
+ //show_image(ai, "aug");
+ //cvWaitKey(0);
+
+ fill_truth_detection(filename, boxes, d.y.vals[i], classes, flip, dx, dy, 1./sx, 1./sy, small_object, w, h);
+
+ cvReleaseImage(&src);
+ }
+ free(random_paths);
+ return d;
+}
+#else // OPENCV
+data load_data_detection(int n, char **paths, int m, int w, int h, int boxes, int classes, int use_flip, float jitter, float hue, float saturation, float exposure, int small_object)
+{
+ char **random_paths = get_random_paths(paths, n, m);
+ int i;
+ data d = { 0 };
+ d.shallow = 0;
+
+ d.X.rows = n;
+ d.X.vals = calloc(d.X.rows, sizeof(float*));
+ d.X.cols = h*w * 3;
+
+ d.y = make_matrix(n, 5 * boxes);
+ 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_strong(-dw, dw);
+ int pright = rand_uniform_strong(-dw, dw);
+ int ptop = rand_uniform_strong(-dh, dh);
+ int pbot = rand_uniform_strong(-dh, dh);
+
+ int swidth = ow - pleft - pright;
+ int sheight = oh - ptop - pbot;
+
+ float sx = (float)swidth / ow;
+ float sy = (float)sheight / oh;
+
+ int flip = use_flip ? random_gen() % 2 : 0;
+ 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);
+ random_distort_image(sized, hue, saturation, exposure);
+ d.X.vals[i] = sized.data;
+
+ fill_truth_detection(random_paths[i], boxes, d.y.vals[i], classes, flip, dx, dy, 1. / sx, 1. / sy, small_object, w, h);
+
+ free_image(orig);
+ free_image(cropped);
+ }
+ free(random_paths);
+ return d;
+}
+#endif // OPENCV
+
+void *load_thread(void *ptr)
+{
+ //srand(time(0));
+ //printf("Loading data: %d\n", random_gen());
+ load_args a = *(struct load_args*)ptr;
+ if(a.exposure == 0) a.exposure = 1;
+ if(a.saturation == 0) a.saturation = 1;
+ if(a.aspect == 0) a.aspect = 1;
+
+ if (a.type == OLD_CLASSIFICATION_DATA){
+ *a.d = load_data_old(a.paths, a.n, a.m, a.labels, a.classes, a.w, a.h);
+ } else if (a.type == CLASSIFICATION_DATA){
+ *a.d = load_data_augment(a.paths, a.n, a.m, a.labels, a.classes, a.hierarchy, a.flip, a.min, a.max, a.size, a.angle, a.aspect, a.hue, a.saturation, a.exposure);
+ } else if (a.type == SUPER_DATA){
+ *a.d = load_data_super(a.paths, a.n, a.m, a.w, a.h, a.scale);
+ } 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, a.hue, a.saturation, a.exposure);
+ } else if (a.type == DETECTION_DATA){
+ *a.d = load_data_detection(a.n, a.paths, a.m, a.w, a.h, a.num_boxes, a.classes, a.flip, a.jitter, a.hue, a.saturation, a.exposure, a.small_object);
+ } 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);
+ }else if (a.type == LETTERBOX_DATA) {
+ *(a.im) = load_image_color(a.path, 0, 0);
+ *(a.resized) = letterbox_image(*(a.im), a.w, a.h);
+ } else if (a.type == TAG_DATA){
+ *a.d = load_data_tag(a.paths, a.n, a.m, a.classes, a.flip, a.min, a.max, a.size, a.angle, a.aspect, a.hue, a.saturation, a.exposure);
+ }
free(ptr);
return 0;
}
-pthread_t load_data_thread(char **paths, int n, int m, char **labels, int k, int w, int h, 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->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");
- }
+ 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;
}
+void *load_threads(void *ptr)
+{
+ //srand(time(0));
+ int i;
+ load_args args = *(load_args *)ptr;
+ if (args.threads == 0) args.threads = 1;
+ data *out = args.d;
+ int total = args.n;
+ free(ptr);
+ data *buffers = calloc(args.threads, sizeof(data));
+ pthread_t *threads = calloc(args.threads, sizeof(pthread_t));
+ for(i = 0; i < args.threads; ++i){
+ args.d = buffers + i;
+ args.n = (i+1) * total/args.threads - i * total/args.threads;
+ threads[i] = load_data_in_thread(args);
+ }
+ for(i = 0; i < args.threads; ++i){
+ pthread_join(threads[i], 0);
+ }
+ *out = concat_datas(buffers, args.threads);
+ out->shallow = 0;
+ for(i = 0; i < args.threads; ++i){
+ buffers[i].shallow = 1;
+ free_data(buffers[i]);
+ }
+ free(buffers);
+ free(threads);
+ return 0;
+}
+
+pthread_t load_data(load_args args)
+{
+ pthread_t thread;
+ struct load_args *ptr = calloc(1, sizeof(struct load_args));
+ *ptr = args;
+ if(pthread_create(&thread, 0, load_threads, 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 = {0};
+ 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_old(char **paths, int n, int m, char **labels, int k, int w, int h)
+{
+ if(m) paths = get_random_paths(paths, n, m);
+ data d = {0};
+ d.shallow = 0;
+ d.X = load_image_paths(paths, n, w, h);
+ d.y = load_labels_paths(paths, n, labels, k, 0);
+ if(m) free(paths);
+ return d;
+}
+
+/*
+ data load_data_study(char **paths, int n, int m, char **labels, int k, int min, int max, int size, float angle, float aspect, float hue, float saturation, float exposure)
+ {
+ data d = {0};
+ d.indexes = calloc(n, sizeof(int));
+ if(m) paths = get_random_paths_indexes(paths, n, m, d.indexes);
+ d.shallow = 0;
+ d.X = load_image_augment_paths(paths, n, flip, min, max, size, angle, aspect, hue, saturation, exposure);
+ d.y = load_labels_paths(paths, n, labels, k);
+ if(m) free(paths);
+ return d;
+ }
+ */
+
+data load_data_super(char **paths, int n, int m, int w, int h, int scale)
+{
+ if(m) paths = get_random_paths(paths, n, m);
+ data d = {0};
+ d.shallow = 0;
+
+ int i;
+ d.X.rows = n;
+ d.X.vals = calloc(n, sizeof(float*));
+ d.X.cols = w*h*3;
+
+ d.y.rows = n;
+ d.y.vals = calloc(n, sizeof(float*));
+ d.y.cols = w*scale * h*scale * 3;
+
+ for(i = 0; i < n; ++i){
+ image im = load_image_color(paths[i], 0, 0);
+ image crop = random_crop_image(im, w*scale, h*scale);
+ int flip = random_gen()%2;
+ if (flip) flip_image(crop);
+ image resize = resize_image(crop, w, h);
+ d.X.vals[i] = resize.data;
+ d.y.vals[i] = crop.data;
+ free_image(im);
+ }
+
+ if(m) free(paths);
+ return d;
+}
+
+data load_data_augment(char **paths, int n, int m, char **labels, int k, tree *hierarchy, int use_flip, int min, int max, int size, float angle, float aspect, float hue, float saturation, float exposure)
+{
+ if(m) paths = get_random_paths(paths, n, m);
+ data d = {0};
+ d.shallow = 0;
+ d.X = load_image_augment_paths(paths, n, use_flip, min, max, size, angle, aspect, hue, saturation, exposure);
+ d.y = load_labels_paths(paths, n, labels, k, hierarchy);
+ if(m) free(paths);
+ return d;
+}
+
+data load_data_tag(char **paths, int n, int m, int k, int use_flip, int min, int max, int size, float angle, float aspect, float hue, float saturation, float exposure)
+{
+ if(m) paths = get_random_paths(paths, n, m);
+ data d = {0};
+ d.w = size;
+ d.h = size;
+ d.shallow = 0;
+ d.X = load_image_augment_paths(paths, n, use_flip, min, max, size, angle, aspect, hue, saturation, exposure);
+ d.y = load_tags_paths(paths, n, k);
+ if(m) free(paths);
+ return d;
+}
+
matrix concat_matrix(matrix m1, matrix m2)
{
int i, count = 0;
@@ -552,16 +1004,28 @@
data concat_data(data d1, data d2)
{
- data d;
+ data d = {0};
d.shallow = 1;
d.X = concat_matrix(d1.X, d2.X);
d.y = concat_matrix(d1.y, d2.y);
return d;
}
+data concat_datas(data *d, int n)
+{
+ int i;
+ data out = {0};
+ for(i = 0; i < n; ++i){
+ data new = concat_data(d[i], out);
+ free_data(out);
+ out = new;
+ }
+ return out;
+}
+
data load_categorical_data_csv(char *filename, int target, int k)
{
- data d;
+ data d = {0};
d.shallow = 0;
matrix X = csv_to_matrix(filename);
float *truth_1d = pop_column(&X, target);
@@ -578,7 +1042,7 @@
data load_cifar10_data(char *filename)
{
- data d;
+ data d = {0};
d.shallow = 0;
long i,j;
matrix X = make_matrix(10000, 3072);
@@ -591,14 +1055,14 @@
for(i = 0; i < 10000; ++i){
unsigned char bytes[3073];
fread(bytes, 1, 3073, fp);
- int class = bytes[0];
- y.vals[i][class] = 1;
+ int class_id = bytes[0];
+ y.vals[i][class_id] = 1;
for(j = 0; j < X.cols; ++j){
X.vals[i][j] = (double)bytes[j+1];
}
}
- translate_data_rows(d, -128);
- scale_data_rows(d, 1./128);
+ //translate_data_rows(d, -128);
+ scale_data_rows(d, 1./255);
//normalize_data_rows(d);
fclose(fp);
return d;
@@ -608,7 +1072,7 @@
{
int j;
for(j = 0; j < n; ++j){
- int index = rand_r(&data_seed)%d.X.rows;
+ int index = random_gen()%d.X.rows;
memcpy(X+j*d.X.cols, d.X.vals[index], d.X.cols*sizeof(float));
memcpy(y+j*d.y.cols, d.y.vals[index], d.y.cols*sizeof(float));
}
@@ -624,10 +1088,21 @@
}
}
+void smooth_data(data d)
+{
+ int i, j;
+ float scale = 1. / d.y.cols;
+ float eps = .1;
+ for(i = 0; i < d.y.rows; ++i){
+ for(j = 0; j < d.y.cols; ++j){
+ d.y.vals[i][j] = eps * scale + (1-eps) * d.y.vals[i][j];
+ }
+ }
+}
data load_all_cifar10()
{
- data d;
+ data d = {0};
d.shallow = 0;
int i,j,b;
matrix X = make_matrix(50000, 3072);
@@ -638,14 +1113,14 @@
for(b = 0; b < 5; ++b){
char buff[256];
- sprintf(buff, "data/cifar10/data_batch_%d.bin", b+1);
+ sprintf(buff, "data/cifar/cifar-10-batches-bin/data_batch_%d.bin", b+1);
FILE *fp = fopen(buff, "rb");
if(!fp) file_error(buff);
for(i = 0; i < 10000; ++i){
unsigned char bytes[3073];
fread(bytes, 1, 3073, fp);
- int class = bytes[0];
- y.vals[i+b*10000][class] = 1;
+ int class_id = bytes[0];
+ y.vals[i+b*10000][class_id] = 1;
for(j = 0; j < X.cols; ++j){
X.vals[i+b*10000][j] = (double)bytes[j+1];
}
@@ -653,16 +1128,64 @@
fclose(fp);
}
//normalize_data_rows(d);
- translate_data_rows(d, -128);
- scale_data_rows(d, 1./128);
+ //translate_data_rows(d, -128);
+ scale_data_rows(d, 1./255);
+ smooth_data(d);
return d;
}
+data load_go(char *filename)
+{
+ FILE *fp = fopen(filename, "rb");
+ matrix X = make_matrix(3363059, 361);
+ matrix y = make_matrix(3363059, 361);
+ int row, col;
+
+ if(!fp) file_error(filename);
+ char *label;
+ int count = 0;
+ while((label = fgetl(fp))){
+ int i;
+ if(count == X.rows){
+ X = resize_matrix(X, count*2);
+ y = resize_matrix(y, count*2);
+ }
+ sscanf(label, "%d %d", &row, &col);
+ char *board = fgetl(fp);
+
+ int index = row*19 + col;
+ y.vals[count][index] = 1;
+
+ for(i = 0; i < 19*19; ++i){
+ float val = 0;
+ if(board[i] == '1') val = 1;
+ else if(board[i] == '2') val = -1;
+ X.vals[count][i] = val;
+ }
+ ++count;
+ free(label);
+ free(board);
+ }
+ X = resize_matrix(X, count);
+ y = resize_matrix(y, count);
+
+ data d = {0};
+ d.shallow = 0;
+ d.X = X;
+ d.y = y;
+
+
+ fclose(fp);
+
+ return d;
+}
+
+
void randomize_data(data d)
{
int i;
for(i = d.X.rows-1; i > 0; --i){
- int index = rand_r(&data_seed)%i;
+ int index = random_gen()%i;
float *swap = d.X.vals[index];
d.X.vals[index] = d.X.vals[i];
d.X.vals[i] = swap;
@@ -697,6 +1220,42 @@
}
}
+data get_data_part(data d, int part, int total)
+{
+ data p = {0};
+ p.shallow = 1;
+ p.X.rows = d.X.rows * (part + 1) / total - d.X.rows * part / total;
+ p.y.rows = d.y.rows * (part + 1) / total - d.y.rows * part / total;
+ p.X.cols = d.X.cols;
+ p.y.cols = d.y.cols;
+ p.X.vals = d.X.vals + d.X.rows * part / total;
+ p.y.vals = d.y.vals + d.y.rows * part / total;
+ return p;
+}
+
+data get_random_data(data d, int num)
+{
+ data r = {0};
+ r.shallow = 1;
+
+ r.X.rows = num;
+ r.y.rows = num;
+
+ r.X.cols = d.X.cols;
+ r.y.cols = d.y.cols;
+
+ r.X.vals = calloc(num, sizeof(float *));
+ r.y.vals = calloc(num, sizeof(float *));
+
+ int i;
+ for(i = 0; i < num; ++i){
+ int index = random_gen()%d.X.rows;
+ r.X.vals[i] = d.X.vals[index];
+ r.y.vals[i] = d.y.vals[index];
+ }
+ return r;
+}
+
data *split_data(data d, int part, int total)
{
data *split = calloc(2, sizeof(data));
--
Gitblit v1.10.0