From 176d65b76583803cf10194c4c70bdc51897f2ae3 Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Mon, 11 Aug 2014 19:52:07 +0000
Subject: [PATCH] Nist NIN testing multi-crop
---
src/network.c | 214 +++++++++++++++++++++++++----------------------------
1 files changed, 101 insertions(+), 113 deletions(-)
diff --git a/src/network.c b/src/network.c
index ef80110..292bba0 100644
--- a/src/network.c
+++ b/src/network.c
@@ -4,11 +4,13 @@
#include "data.h"
#include "utils.h"
+#include "crop_layer.h"
#include "connected_layer.h"
#include "convolutional_layer.h"
#include "maxpool_layer.h"
#include "normalization_layer.h"
#include "softmax_layer.h"
+#include "dropout_layer.h"
network make_network(int n, int batch)
{
@@ -25,94 +27,6 @@
return net;
}
-void print_convolutional_cfg(FILE *fp, convolutional_layer *l, int first)
-{
- int i;
- fprintf(fp, "[convolutional]\n");
- if(first) fprintf(fp, "batch=%d\n"
- "height=%d\n"
- "width=%d\n"
- "channels=%d\n",
- l->batch,l->h, l->w, l->c);
- fprintf(fp, "filters=%d\n"
- "size=%d\n"
- "stride=%d\n"
- "activation=%s\n",
- l->n, l->size, l->stride,
- get_activation_string(l->activation));
- fprintf(fp, "data=");
- for(i = 0; i < l->n; ++i) fprintf(fp, "%g,", l->biases[i]);
- for(i = 0; i < l->n*l->c*l->size*l->size; ++i) fprintf(fp, "%g,", l->filters[i]);
- fprintf(fp, "\n\n");
-}
-void print_connected_cfg(FILE *fp, connected_layer *l, int first)
-{
- int i;
- fprintf(fp, "[connected]\n");
- if(first) fprintf(fp, "batch=%d\ninput=%d\n", l->batch, l->inputs);
- fprintf(fp, "output=%d\n"
- "activation=%s\n",
- l->outputs,
- get_activation_string(l->activation));
- fprintf(fp, "data=");
- for(i = 0; i < l->outputs; ++i) fprintf(fp, "%g,", l->biases[i]);
- for(i = 0; i < l->inputs*l->outputs; ++i) fprintf(fp, "%g,", l->weights[i]);
- fprintf(fp, "\n\n");
-}
-
-void print_maxpool_cfg(FILE *fp, maxpool_layer *l, int first)
-{
- fprintf(fp, "[maxpool]\n");
- if(first) fprintf(fp, "batch=%d\n"
- "height=%d\n"
- "width=%d\n"
- "channels=%d\n",
- l->batch,l->h, l->w, l->c);
- fprintf(fp, "stride=%d\n\n", l->stride);
-}
-
-void print_normalization_cfg(FILE *fp, normalization_layer *l, int first)
-{
- fprintf(fp, "[localresponsenormalization]\n");
- if(first) fprintf(fp, "batch=%d\n"
- "height=%d\n"
- "width=%d\n"
- "channels=%d\n",
- l->batch,l->h, l->w, l->c);
- fprintf(fp, "size=%d\n"
- "alpha=%g\n"
- "beta=%g\n"
- "kappa=%g\n\n", l->size, l->alpha, l->beta, l->kappa);
-}
-
-void print_softmax_cfg(FILE *fp, softmax_layer *l, int first)
-{
- fprintf(fp, "[softmax]\n");
- if(first) fprintf(fp, "batch=%d\ninput=%d\n", l->batch, l->inputs);
- fprintf(fp, "\n");
-}
-
-void save_network(network net, char *filename)
-{
- FILE *fp = fopen(filename, "w");
- if(!fp) file_error(filename);
- int i;
- for(i = 0; i < net.n; ++i)
- {
- if(net.types[i] == CONVOLUTIONAL)
- print_convolutional_cfg(fp, (convolutional_layer *)net.layers[i], i==0);
- else if(net.types[i] == CONNECTED)
- print_connected_cfg(fp, (connected_layer *)net.layers[i], i==0);
- else if(net.types[i] == MAXPOOL)
- print_maxpool_cfg(fp, (maxpool_layer *)net.layers[i], i==0);
- else if(net.types[i] == NORMALIZATION)
- print_normalization_cfg(fp, (normalization_layer *)net.layers[i], i==0);
- else if(net.types[i] == SOFTMAX)
- print_softmax_cfg(fp, (softmax_layer *)net.layers[i], i==0);
- }
- fclose(fp);
-}
-
#ifdef GPU
void forward_network(network net, float *input, int train)
{
@@ -143,6 +57,11 @@
forward_softmax_layer(layer, input);
input = layer.output;
}
+ else if(net.types[i] == CROP){
+ crop_layer layer = *(crop_layer *)net.layers[i];
+ forward_crop_layer(layer, input);
+ input = layer.output;
+ }
else if(net.types[i] == MAXPOOL){
maxpool_layer layer = *(maxpool_layer *)net.layers[i];
forward_maxpool_layer(layer, input);
@@ -169,7 +88,12 @@
}
else if(net.types[i] == CONNECTED){
connected_layer layer = *(connected_layer *)net.layers[i];
- forward_connected_layer(layer, input, train);
+ forward_connected_layer(layer, input);
+ input = layer.output;
+ }
+ else if(net.types[i] == CROP){
+ crop_layer layer = *(crop_layer *)net.layers[i];
+ forward_crop_layer(layer, input);
input = layer.output;
}
else if(net.types[i] == SOFTMAX){
@@ -187,17 +111,22 @@
forward_normalization_layer(layer, input);
input = layer.output;
}
+ else if(net.types[i] == DROPOUT){
+ if(!train) continue;
+ dropout_layer layer = *(dropout_layer *)net.layers[i];
+ forward_dropout_layer(layer, input);
+ }
}
}
#endif
-void update_network(network net, float step, float momentum, float decay)
+void update_network(network net)
{
int i;
for(i = 0; i < net.n; ++i){
if(net.types[i] == CONVOLUTIONAL){
convolutional_layer layer = *(convolutional_layer *)net.layers[i];
- update_convolutional_layer(layer, step, momentum, decay);
+ update_convolutional_layer(layer);
}
else if(net.types[i] == MAXPOOL){
//maxpool_layer layer = *(maxpool_layer *)net.layers[i];
@@ -210,7 +139,7 @@
}
else if(net.types[i] == CONNECTED){
connected_layer layer = *(connected_layer *)net.layers[i];
- update_connected_layer(layer, step, momentum, decay);
+ update_connected_layer(layer);
}
}
}
@@ -226,6 +155,8 @@
} else if(net.types[i] == SOFTMAX){
softmax_layer layer = *(softmax_layer *)net.layers[i];
return layer.output;
+ } else if(net.types[i] == DROPOUT){
+ return get_network_output_layer(net, i-1);
} else if(net.types[i] == CONNECTED){
connected_layer layer = *(connected_layer *)net.layers[i];
return layer.output;
@@ -251,6 +182,8 @@
} else if(net.types[i] == SOFTMAX){
softmax_layer layer = *(softmax_layer *)net.layers[i];
return layer.delta;
+ } else if(net.types[i] == DROPOUT){
+ return get_network_delta_layer(net, i-1);
} else if(net.types[i] == CONNECTED){
connected_layer layer = *(connected_layer *)net.layers[i];
return layer.delta;
@@ -272,7 +205,9 @@
for(i = 0; i < get_network_output_size(net)*net.batch; ++i){
//if(i %get_network_output_size(net) == 0) printf("\n");
//printf("%5.2f %5.2f, ", out[i], truth[i]);
+ //if(i == get_network_output_size(net)) printf("\n");
delta[i] = truth[i] - out[i];
+ //printf("%.10f, ", out[i]);
sum += delta[i]*delta[i];
}
//printf("\n");
@@ -324,17 +259,17 @@
return error;
}
-float train_network_datum(network net, float *x, float *y, float step, float momentum, float decay)
+float train_network_datum(network net, float *x, float *y)
{
forward_network(net, x, 1);
//int class = get_predicted_class_network(net);
float error = backward_network(net, x, y);
- update_network(net, step, momentum, decay);
+ update_network(net);
//return (y[class]?1:0);
return error;
}
-float train_network_sgd(network net, data d, int n, float step, float momentum,float decay)
+float train_network_sgd(network net, data d, int n)
{
int batch = net.batch;
float *X = calloc(batch*d.X.cols, sizeof(float));
@@ -342,15 +277,17 @@
int i,j;
float sum = 0;
+ int index = 0;
for(i = 0; i < n; ++i){
for(j = 0; j < batch; ++j){
- int index = rand()%d.X.rows;
+ index = rand()%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));
}
- float err = train_network_datum(net, X, y, step, momentum, decay);
+
+ float err = train_network_datum(net, X, y);
sum += err;
- //train_network_datum(net, X, y, step, momentum, decay);
+ //train_network_datum(net, X, y);
/*
float *y = d.y.vals[index];
int class = get_predicted_class_network(net);
@@ -376,35 +313,36 @@
//}
}
//printf("Accuracy: %f\n",(float) correct/n);
+ //show_image(float_to_image(32,32,3,X), "Orig");
free(X);
free(y);
return (float)sum/(n*batch);
}
-float train_network_batch(network net, data d, int n, float step, float momentum,float decay)
+float train_network_batch(network net, data d, int n)
{
- int i;
- int correct = 0;
+ int i,j;
+ float sum = 0;
+ int batch = 2;
for(i = 0; i < n; ++i){
- int index = rand()%d.X.rows;
- float *x = d.X.vals[index];
- float *y = d.y.vals[index];
- forward_network(net, x, 1);
- int class = get_predicted_class_network(net);
- backward_network(net, x, y);
- correct += (y[class]?1:0);
+ for(j = 0; j < batch; ++j){
+ int index = rand()%d.X.rows;
+ float *x = d.X.vals[index];
+ float *y = d.y.vals[index];
+ forward_network(net, x, 1);
+ sum += backward_network(net, x, y);
+ }
+ update_network(net);
}
- update_network(net, step, momentum, decay);
- return (float)correct/n;
-
+ return (float)sum/(n*batch);
}
-void train_network(network net, data d, float step, float momentum, float decay)
+void train_network(network net, data d)
{
int i;
int correct = 0;
for(i = 0; i < d.X.rows; ++i){
- correct += train_network_datum(net, d.X.vals[i], d.y.vals[i], step, momentum, decay);
+ correct += train_network_datum(net, d.X.vals[i], d.y.vals[i]);
if(i%100 == 0){
visualize_network(net);
cvWaitKey(10);
@@ -428,6 +366,9 @@
else if(net.types[i] == CONNECTED){
connected_layer layer = *(connected_layer *)net.layers[i];
return layer.inputs;
+ } else if(net.types[i] == DROPOUT){
+ dropout_layer layer = *(dropout_layer *) net.layers[i];
+ return layer.inputs;
}
else if(net.types[i] == SOFTMAX){
softmax_layer layer = *(softmax_layer *)net.layers[i];
@@ -451,6 +392,9 @@
else if(net.types[i] == CONNECTED){
connected_layer layer = *(connected_layer *)net.layers[i];
return layer.outputs;
+ } else if(net.types[i] == DROPOUT){
+ dropout_layer layer = *(dropout_layer *) net.layers[i];
+ return layer.inputs;
}
else if(net.types[i] == SOFTMAX){
softmax_layer layer = *(softmax_layer *)net.layers[i];
@@ -516,6 +460,10 @@
normalization_layer layer = *(normalization_layer *)net.layers[i];
return get_normalization_image(layer);
}
+ else if(net.types[i] == CROP){
+ crop_layer layer = *(crop_layer *)net.layers[i];
+ return get_crop_image(layer);
+ }
return make_empty_image(0,0,0);
}
@@ -534,6 +482,7 @@
image *prev = 0;
int i;
char buff[256];
+ show_image(get_network_image_layer(net, 0), "Crop");
for(i = 0; i < net.n; ++i){
sprintf(buff, "Layer %d", i);
if(net.types[i] == CONVOLUTIONAL){
@@ -554,6 +503,31 @@
return out;
}
+matrix network_predict_data_multi(network net, data test, int n)
+{
+ int i,j,b,m;
+ int k = get_network_output_size(net);
+ matrix pred = make_matrix(test.X.rows, k);
+ float *X = calloc(net.batch*test.X.rows, sizeof(float));
+ for(i = 0; i < test.X.rows; i += net.batch){
+ for(b = 0; b < net.batch; ++b){
+ if(i+b == test.X.rows) break;
+ memcpy(X+b*test.X.cols, test.X.vals[i+b], test.X.cols*sizeof(float));
+ }
+ for(m = 0; m < n; ++m){
+ float *out = network_predict(net, X);
+ for(b = 0; b < net.batch; ++b){
+ if(i+b == test.X.rows) break;
+ for(j = 0; j < k; ++j){
+ pred.vals[i+b][j] += out[j+b*k]/n;
+ }
+ }
+ }
+ }
+ free(X);
+ return pred;
+}
+
matrix network_predict_data(network net, data test)
{
int i,j,b;
@@ -595,6 +569,12 @@
image m = get_maxpool_image(layer);
n = m.h*m.w*m.c;
}
+ else if(net.types[i] == CROP){
+ crop_layer layer = *(crop_layer *)net.layers[i];
+ output = layer.output;
+ image m = get_crop_image(layer);
+ n = m.h*m.w*m.c;
+ }
else if(net.types[i] == CONNECTED){
connected_layer layer = *(connected_layer *)net.layers[i];
output = layer.output;
@@ -623,4 +603,12 @@
return acc;
}
+float network_accuracy_multi(network net, data d, int n)
+{
+ matrix guess = network_predict_data_multi(net, d, n);
+ float acc = matrix_accuracy(d.y, guess);
+ free_matrix(guess);
+ return acc;
+}
+
--
Gitblit v1.10.0