diff --git a/src/coco.c b/src/coco.c
index 3f74be7d07f7e0ca03bed58d7aca852349872be3..ed53cef4cf55212ae5203ed33f55f73eec0fca0b 100644
--- a/src/coco.c
+++ b/src/coco.c
@@ -1,3 +1,5 @@
+#include <stdio.h>
+
#include "network.h"
#include "detection_layer.h"
#include "cost_layer.h"
@@ -8,6 +10,8 @@
char *coco_classes[] = {"person","bicycle","car","motorcycle","airplane","bus","train","truck","boat","traffic light","fire hydrant","stop sign","parking meter","bench","bird","cat","dog","horse","sheep","cow","elephant","bear","zebra","giraffe","backpack","umbrella","handbag","tie","suitcase","frisbee","skis","snowboard","sports ball","kite","baseball bat","baseball glove","skateboard","surfboard","tennis racket","bottle","wine glass","cup","fork","knife","spoon","bowl","banana","apple","sandwich","orange","broccoli","carrot","hot dog","pizza","donut","cake","chair","couch","potted plant","bed","dining table","toilet","tv","laptop","mouse","remote","keyboard","cell phone","microwave","oven","toaster","sink","refrigerator","book","clock","vase","scissors","teddy bear","hair drier","toothbrush"};
+int coco_ids[] = {1,2,3,4,5,6,7,8,9,10,11,13,14,15,16,17,18,19,20,21,22,23,24,25,27,28,31,32,33,34,35,36,37,38,39,40,41,42,43,44,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,67,70,72,73,74,75,76,77,78,79,80,81,82,84,85,86,87,88,89,90};
+
void draw_coco(image im, float *box, int side, int objectness, char *label)
{
int classes = 80;
@@ -144,7 +148,7 @@ void convert_cocos(float *predictions, int classes, int objectness, int backgrou
}
}
-void print_cocos(FILE **fps, char *id, box *boxes, float **probs, int num_boxes, int classes, int w, int h)
+void print_cocos(FILE *fp, int image_id, box *boxes, float **probs, int num_boxes, int classes, int w, int h)
{
int i, j;
for(i = 0; i < num_boxes*num_boxes; ++i){
@@ -158,13 +162,23 @@ void print_cocos(FILE **fps, char *id, box *boxes, float **probs, int num_boxes,
if (xmax > w) xmax = w;
if (ymax > h) ymax = h;
+ float bx = xmin;
+ float by = ymin;
+ float bw = xmax - xmin;
+ float bh = ymax - ymin;
+
for(j = 0; j < classes; ++j){
- if (probs[i][j]) fprintf(fps[j], "%s %f %f %f %f %f\n", id, probs[i][j],
- xmin, ymin, xmax, ymax);
+ if (probs[i][j]) fprintf(fp, "{\"image_id\":%d, \"category_id\":%d, \"bbox\":[%f, %f, %f, %f], \"score\":%f},\n", image_id, coco_ids[j], bx, by, bw, bh, probs[i][j]);
}
}
}
+int get_coco_image_id(char *filename)
+{
+ char *p = strrchr(filename, '_');
+ return atoi(p+1);
+}
+
void validate_coco(char *cfgfile, char *weightfile)
{
network net = parse_network_cfg(cfgfile);
@@ -176,8 +190,8 @@ void validate_coco(char *cfgfile, char *weightfile)
fprintf(stderr, "Learning Rate: %g, Momentum: %g, Decay: %g\n", net.learning_rate, net.momentum, net.decay);
srand(time(0));
- char *base = "results/comp4_det_test_";
- list *plist = get_paths("data/voc.2012test.list");
+ char *base = "/home/pjreddie/backup/";
+ list *plist = get_paths("data/coco_val_5k.list");
char **paths = (char **)list_to_array(plist);
int classes = layer.classes;
@@ -186,12 +200,11 @@ void validate_coco(char *cfgfile, char *weightfile)
int num_boxes = sqrt(get_detection_layer_locations(layer));
int j;
- FILE **fps = calloc(classes, sizeof(FILE *));
- for(j = 0; j < classes; ++j){
- char buff[1024];
- snprintf(buff, 1024, "%s%s.txt", base, coco_classes[j]);
- fps[j] = fopen(buff, "w");
- }
+ char buff[1024];
+ snprintf(buff, 1024, "%s/coco_results.json", base);
+ FILE *fp = fopen(buff, "w");
+ fprintf(fp, "[\n");
+
box *boxes = calloc(num_boxes*num_boxes, sizeof(box));
float **probs = calloc(num_boxes*num_boxes, sizeof(float *));
for(j = 0; j < num_boxes*num_boxes; ++j) probs[j] = calloc(classes, sizeof(float *));
@@ -200,7 +213,7 @@ void validate_coco(char *cfgfile, char *weightfile)
int i=0;
int t;
- float thresh = .001;
+ float thresh = .01;
int nms = 1;
float iou_thresh = .5;
@@ -226,19 +239,21 @@ void validate_coco(char *cfgfile, char *weightfile)
}
for(t = 0; t < nthreads && i+t-nthreads < m; ++t){
char *path = paths[i+t-nthreads];
- char *id = basecfg(path);
+ int image_id = get_coco_image_id(path);
float *X = val_resized[t].data;
float *predictions = network_predict(net, X);
int w = val[t].w;
int h = val[t].h;
convert_cocos(predictions, classes, objectness, background, num_boxes, w, h, thresh, probs, boxes);
if (nms) do_nms(boxes, probs, num_boxes, classes, iou_thresh);
- print_cocos(fps, id, boxes, probs, num_boxes, classes, w, h);
- free(id);
+ print_cocos(fp, image_id, boxes, probs, num_boxes, classes, w, h);
free_image(val[t]);
free_image(val_resized[t]);
}
}
+ fseek(fp, -2, SEEK_CUR);
+ fprintf(fp, "\n]\n");
+ fclose(fp);
fprintf(stderr, "Total Detection Time: %f Seconds\n", (double)(time(0) - start));
}
diff --git a/src/crop_layer_kernels.cu b/src/crop_layer_kernels.cu
index 2c3aefed6a5abcf95f7db1cb66f65da7d78ceeb5..fc7fcbdb9b36fc492b5bc64d5f59418412dddf74 100644
--- a/src/crop_layer_kernels.cu
+++ b/src/crop_layer_kernels.cu
@@ -5,8 +5,6 @@ extern "C" {
#include "image.h"
}
-#define BLOCK 256
-
__device__ float get_pixel_kernel(float *image, int w, int h, int x, int y, int c)
{
if(x < 0 || x >= w || y < 0 || y >= h) return 0;
diff --git a/src/cuda.h b/src/cuda.h
index ff0d5f337bfb5cce4c1dafb9f2fc55720d012896..9d949b0da25f9dc911ce16ca505f701ea300cb75 100644
--- a/src/cuda.h
+++ b/src/cuda.h
@@ -5,7 +5,7 @@ extern int gpu_index;
#ifdef GPU
-#define BLOCK 256
+#define BLOCK 512
#include "cuda_runtime.h"
#include "curand.h"
diff --git a/src/detection.c b/src/detection.c
index 55c75dee75710b13b8f8063b118bf7613fdcee1a..f5957010ad9dcf4ac8a076fc6db263353ee0c9bb 100644
--- a/src/detection.c
+++ b/src/detection.c
@@ -79,7 +79,7 @@ void train_detection(char *cfgfile, char *weightfile)
paths = (char **)list_to_array(plist);
pthread_t load_thread = load_data_detection_thread(imgs, paths, plist->size, classes, net.w, net.h, side, side, background, &buffer);
clock_t time;
- while(i*imgs < N*120){
+ while(i*imgs < N*130){
i += 1;
time=clock();
pthread_join(load_thread, 0);
@@ -95,7 +95,7 @@ void train_detection(char *cfgfile, char *weightfile)
printf("%d: %f, %f avg, %lf seconds, %d images\n", i, loss, avg_loss, sec(clock()-time), i*imgs);
if((i-1)*imgs <= N && i*imgs > N){
- fprintf(stderr, "Starting second stage...\n");
+ fprintf(stderr, "First stage done\n");
net.learning_rate *= 10;
char buff[256];
sprintf(buff, "%s/%s_first_stage.weights", backup_directory, base);
@@ -109,6 +109,13 @@ void train_detection(char *cfgfile, char *weightfile)
save_weights(net, buff);
return;
}
+ if((i-1)*imgs <= 120*N && i*imgs > N*120){
+ fprintf(stderr, "Third stage done.\n");
+ char buff[256];
+ sprintf(buff, "%s/%s_third_stage.weights", backup_directory, base);
+ net.layers[net.n-1].rescore = 1;
+ save_weights(net, buff);
+ }
if(i%1000==0){
char buff[256];
sprintf(buff, "%s/%s_%d.weights", backup_directory, base, i);
@@ -176,7 +183,7 @@ void validate_detection(char *cfgfile, char *weightfile)
srand(time(0));
char *base = "results/comp4_det_test_";
- list *plist = get_paths("data/voc.2012test.list");
+ list *plist = get_paths("/home/pjreddie/data/voc/test/2007_test.txt");
char **paths = (char **)list_to_array(plist);
int classes = layer.classes;
diff --git a/src/softmax_layer_kernels.cu b/src/softmax_layer_kernels.cu
index 8fbaf1981dbd44c74ac00f67ab9de2ad6f741bb9..7e1338703346bd7638d80f40d420792052d58da1 100644
--- a/src/softmax_layer_kernels.cu
+++ b/src/softmax_layer_kernels.cu
@@ -4,8 +4,6 @@ extern "C" {
#include "blas.h"
}
-#define BLOCK 256
-
__global__ void forward_softmax_layer_kernel(int n, int batch, float *input, float *output)
{
int b = (blockIdx.x + blockIdx.y*gridDim.x) * blockDim.x + threadIdx.x;