diff --git a/src/axpy.cl b/src/axpy.cl
index 901a826637b2424a1a8a36985cce5c6fa84851c7..396389d98c4ae60663ed3820468af72e15ff338b 100644
--- a/src/axpy.cl
+++ b/src/axpy.cl
@@ -10,6 +10,12 @@ __kernel void scal(int N, float ALPHA, __global float *X, int INCX)
X[i*INCX] *= ALPHA;
}
+__kernel void mask(int n, __global float *x, __global float *mask, int mod)
+{
+ int i = get_global_id(0);
+ x[i] = (mask[(i/mod)*mod]) ? x[i] : 0;
+}
+
__kernel void copy(int N, __global float *X, int OFFX, int INCX, __global float *Y, int OFFY, int INCY)
{
int i = get_global_id(0);
diff --git a/src/cnn.c b/src/cnn.c
index 29f9565210c9dd4530ff8d85aa8dc2bacb0699bb..8a4899c53b74d794e7d6b3ee88fe9b8cf5cd547a 100644
--- a/src/cnn.c
+++ b/src/cnn.c
@@ -314,15 +314,14 @@ void train_detection_net()
int imgs = 1000/net.batch+1;
srand(time(0));
int i = 0;
- char **labels = get_labels("/home/pjreddie/data/imagenet/cls.labels.list");
- list *plist = get_paths("/data/imagenet/cls.train.list");
+ list *plist = get_paths("/home/pjreddie/data/imagenet/horse.txt");
char **paths = (char **)list_to_array(plist);
printf("%d\n", plist->size);
clock_t time;
while(1){
i += 1;
time=clock();
- data train = load_data_random(imgs*net.batch, paths, plist->size, labels, 1000, 256, 256);
+ data train = load_data_detection_random(imgs*net.batch, paths, plist->size, 256, 256, 8, 8, 256);
//translate_data_rows(train, -144);
normalize_data_rows(train);
printf("Loaded: %lf seconds\n", sec(clock()-time));
@@ -346,7 +345,7 @@ void train_imagenet()
{
float avg_loss = 1;
//network net = parse_network_cfg("/home/pjreddie/imagenet_backup/alexnet_1270.cfg");
- network net = parse_network_cfg("cfg/alexnet.cfg");
+ network net = parse_network_cfg("cfg/trained_alexnet.cfg");
printf("Learning Rate: %g, Momentum: %g, Decay: %g\n", net.learning_rate, net.momentum, net.decay);
int imgs = 1000/net.batch+1;
srand(time(0));
@@ -412,6 +411,29 @@ void validate_imagenet(char *filename)
}
}
+void test_detection()
+{
+ network net = parse_network_cfg("cfg/detnet_test.cfg");
+ //imgs=1;
+ srand(2222222);
+ int i = 0;
+ clock_t time;
+ char filename[256];
+ int indexes[10];
+ while(1){
+ fgets(filename, 256, stdin);
+ image im = load_image_color(filename, 256, 256);
+ z_normalize_image(im);
+ printf("%d %d %d\n", im.h, im.w, im.c);
+ float *X = im.data;
+ time=clock();
+ float *predictions = network_predict(net, X);
+ top_predictions(net, 10, indexes);
+ printf("%s: Predicted in %f seconds.\n", filename, sec(clock()-time));
+ free_image(im);
+ }
+}
+
void test_imagenet()
{
network net = parse_network_cfg("cfg/imagenet_test.cfg");
@@ -717,6 +739,7 @@ int main(int argc, char *argv[])
return 0;
}
if(0==strcmp(argv[1], "train")) train_imagenet();
+ else if(0==strcmp(argv[1], "detection")) train_detection_net();
else if(0==strcmp(argv[1], "asirra")) train_asirra();
else if(0==strcmp(argv[1], "nist")) train_nist();
else if(0==strcmp(argv[1], "test_correct")) test_gpu_net();
@@ -726,7 +749,6 @@ int main(int argc, char *argv[])
#ifdef GPU
else if(0==strcmp(argv[1], "test_gpu")) test_gpu_blas();
#endif
- test_parser();
fprintf(stderr, "Success!\n");
return 0;
}
diff --git a/src/cost_layer.c b/src/cost_layer.c
index 1df0ed4f527837f233fb3850aec43ed0d6d96c2c..6614b94896f32feff15d6cb5a99b07b52a80ad47 100644
--- a/src/cost_layer.c
+++ b/src/cost_layer.c
@@ -2,15 +2,36 @@
#include "utils.h"
#include "mini_blas.h"
#include <math.h>
+#include <string.h>
#include <stdlib.h>
#include <stdio.h>
-cost_layer *make_cost_layer(int batch, int inputs)
+COST_TYPE get_cost_type(char *s)
+{
+ if (strcmp(s, "sse")==0) return SSE;
+ if (strcmp(s, "detection")==0) return DETECTION;
+ fprintf(stderr, "Couldn't find activation function %s, going with SSE\n", s);
+ return SSE;
+}
+
+char *get_cost_string(COST_TYPE a)
+{
+ switch(a){
+ case SSE:
+ return "sse";
+ case DETECTION:
+ return "detection";
+ }
+ return "sse";
+}
+
+cost_layer *make_cost_layer(int batch, int inputs, COST_TYPE type)
{
fprintf(stderr, "Cost Layer: %d inputs\n", inputs);
cost_layer *layer = calloc(1, sizeof(cost_layer));
layer->batch = batch;
layer->inputs = inputs;
+ layer->type = type;
layer->delta = calloc(inputs*batch, sizeof(float));
layer->output = calloc(1, sizeof(float));
#ifdef GPU
@@ -24,6 +45,12 @@ void forward_cost_layer(cost_layer layer, float *input, float *truth)
if (!truth) return;
copy_cpu(layer.batch*layer.inputs, truth, 1, layer.delta, 1);
axpy_cpu(layer.batch*layer.inputs, -1, input, 1, layer.delta, 1);
+ if(layer.type == DETECTION){
+ int i;
+ for(i = 0; i < layer.batch*layer.inputs; ++i){
+ if((i%5) && !truth[(i/5)*5]) layer.delta[i] = 0;
+ }
+ }
*(layer.output) = dot_cpu(layer.batch*layer.inputs, layer.delta, 1, layer.delta, 1);
}
@@ -33,6 +60,38 @@ void backward_cost_layer(const cost_layer layer, float *input, float *delta)
}
#ifdef GPU
+
+cl_kernel get_mask_kernel()
+{
+ static int init = 0;
+ static cl_kernel kernel;
+ if(!init){
+ kernel = get_kernel("src/axpy.cl", "mask", 0);
+ init = 1;
+ }
+ return kernel;
+}
+
+void mask_ongpu(int n, cl_mem x, cl_mem mask, int mod)
+{
+ cl_setup();
+ cl_kernel kernel = get_mask_kernel();
+ cl_command_queue queue = cl.queue;
+
+ cl_uint i = 0;
+ cl.error = clSetKernelArg(kernel, i++, sizeof(n), (void*) &n);
+ cl.error = clSetKernelArg(kernel, i++, sizeof(x), (void*) &x);
+ cl.error = clSetKernelArg(kernel, i++, sizeof(mask), (void*) &mask);
+ cl.error = clSetKernelArg(kernel, i++, sizeof(mod), (void*) &mod);
+ check_error(cl);
+
+ const size_t global_size[] = {n};
+
+ cl.error = clEnqueueNDRangeKernel(queue, kernel, 1, 0, global_size, 0, 0, 0, 0);
+ check_error(cl);
+
+}
+
void forward_cost_layer_gpu(cost_layer layer, cl_mem input, cl_mem truth)
{
if (!truth) return;
@@ -40,6 +99,10 @@ void forward_cost_layer_gpu(cost_layer layer, cl_mem input, cl_mem truth)
copy_ongpu(layer.batch*layer.inputs, truth, 1, layer.delta_cl, 1);
axpy_ongpu(layer.batch*layer.inputs, -1, input, 1, layer.delta_cl, 1);
+ if(layer.type==DETECTION){
+ mask_ongpu(layer.inputs*layer.batch, layer.delta_cl, truth, 5);
+ }
+
cl_read_array(layer.delta_cl, layer.delta, layer.batch*layer.inputs);
*(layer.output) = dot_cpu(layer.batch*layer.inputs, layer.delta, 1, layer.delta, 1);
//printf("%f\n", *layer.output);
diff --git a/src/cost_layer.h b/src/cost_layer.h
index edda8f95970bc25a681e5f8fbee9ffe00a3a3b2d..2f1cd5510521bb4c1b4cfef0a633ced80dee2c5f 100644
--- a/src/cost_layer.h
+++ b/src/cost_layer.h
@@ -2,17 +2,24 @@
#define COST_LAYER_H
#include "opencl.h"
+typedef enum{
+ SSE, DETECTION
+} COST_TYPE;
+
typedef struct {
int inputs;
int batch;
float *delta;
float *output;
+ COST_TYPE type;
#ifdef GPU
cl_mem delta_cl;
#endif
} cost_layer;
-cost_layer *make_cost_layer(int batch, int inputs);
+COST_TYPE get_cost_type(char *s);
+char *get_cost_string(COST_TYPE a);
+cost_layer *make_cost_layer(int batch, int inputs, COST_TYPE type);
void forward_cost_layer(const cost_layer layer, float *input, float *truth);
void backward_cost_layer(const cost_layer layer, float *input, float *delta);
diff --git a/src/data.c b/src/data.c
index 9b57391c2bec49c0cbdf5ea4ae5a63ce16404d71..3627fcba5767280c302bb421da7b447284e52812 100644
--- a/src/data.c
+++ b/src/data.c
@@ -26,6 +26,7 @@ void fill_truth_detection(char *path, float *truth, int height, int width, int n
char *labelpath = find_replace(path, "imgs", "det");
labelpath = find_replace(labelpath, ".JPEG", ".txt");
FILE *file = fopen(labelpath, "r");
+ if(!file) file_error(labelpath);
int x, y, h, w;
while(fscanf(file, "%d %d %d %d", &x, &y, &w, &h) == 4){
int i = x/box_width;
@@ -34,6 +35,7 @@ void fill_truth_detection(char *path, float *truth, int height, int width, int n
float dw = (float)(y%box_width)/box_width;
float sh = h/scale;
float sw = w/scale;
+ //printf("%d %d %f %f\n", i, j, dh, dw);
int index = (i+j*num_width)*5;
truth[index++] = 1;
truth[index++] = dh;
@@ -41,6 +43,7 @@ void fill_truth_detection(char *path, float *truth, int height, int width, int n
truth[index++] = sh;
truth[index++] = sw;
}
+ fclose(file);
}
void fill_truth(char *path, char **labels, int k, float *truth)
@@ -125,7 +128,7 @@ void free_data(data d)
}
}
-data load_data_detection_random(int n, char **paths, int m, char **labels, int h, int w, int nh, int nw, float scale)
+data load_data_detection_random(int n, char **paths, int m, int h, int w, int nh, int nw, float scale)
{
char **random_paths = calloc(n, sizeof(char*));
int i;
diff --git a/src/data.h b/src/data.h
index 366a382b8525192e9a193dafdeb4a1ff65037427..b9707c966fb5670be4010966e4e1f71b9fd85400 100644
--- a/src/data.h
+++ b/src/data.h
@@ -14,7 +14,7 @@ typedef struct{
void free_data(data d);
data load_data(char **paths, int n, char **labels, int k, int h, int w);
data load_data_random(int n, char **paths, int m, char **labels, int k, int h, int w);
-data load_data_detection_random(int n, char **paths, int m, char **labels, int h, int w, int nh, int nw, float scale);
+data load_data_detection_random(int n, char **paths, int m, int h, int w, int nh, int nw, float scale);
data load_data_image_pathfile(char *filename, char **labels, int k, int h, int w);
data load_cifar10_data(char *filename);
data load_all_cifar10();
diff --git a/src/parser.c b/src/parser.c
index 79d4a3a99f04533b30b8c50f70120d09466608e1..206975395dbb36f7e1bc675fa60c27bac1d1655c 100644
--- a/src/parser.c
+++ b/src/parser.c
@@ -165,7 +165,9 @@ cost_layer *parse_cost(list *options, network *net, int count)
}else{
input = get_network_output_size_layer(*net, count-1);
}
- cost_layer *layer = make_cost_layer(net->batch, input);
+ char *type_s = option_find_str(options, "type", "sse");
+ COST_TYPE type = get_cost_type(type_s);
+ cost_layer *layer = make_cost_layer(net->batch, input, type);
option_unused(options);
return layer;
}
@@ -565,7 +567,7 @@ void print_softmax_cfg(FILE *fp, softmax_layer *l, network net, int count)
void print_cost_cfg(FILE *fp, cost_layer *l, network net, int count)
{
- fprintf(fp, "[cost]\n");
+ fprintf(fp, "[cost]\ntype=%s\n", get_cost_string(l->type));
if(count == 0) fprintf(fp, "batch=%d\ninput=%d\n", l->batch, l->inputs);
fprintf(fp, "\n");
}