so much need to commit

Makefile

@@ -1,5 +1,5 @@
-GPU=1
-OPENCV=1
+GPU=0
+OPENCV=0
 DEBUG=0
 
 ARCH= --gpu-architecture=compute_20 --gpu-code=compute_20 
@@ -11,7 +11,7 @@ OBJDIR=./obj/
 CC=gcc
 NVCC=nvcc
 OPTS=-Ofast
-LDFLAGS= -lm -pthread -lstdc++ 
+LDFLAGS= -lm -pthread 
 COMMON= 
 CFLAGS=-Wall -Wfatal-errors 
 
@@ -34,8 +34,9 @@ CFLAGS+= -DGPU
 LDFLAGS+= -L/usr/local/cuda/lib64 -lcuda -lcudart -lcublas -lcurand
 endif
 
-OBJ=gemm.o utils.o cuda.o deconvolutional_layer.o convolutional_layer.o list.o image.o activations.o im2col.o col2im.o blas.o crop_layer.o dropout_layer.o maxpool_layer.o softmax_layer.o data.o matrix.o network.o connected_layer.o cost_layer.o parser.o option_list.o darknet.o detection_layer.o imagenet.o captcha.o route_layer.o writing.o box.o nightmare.o normalization_layer.o avgpool_layer.o coco.o dice.o yolo.o layer.o compare.o classifier.o local_layer.o swag.o shortcut_layer.o activation_layer.o rnn_layer.o rnn.o rnn_vid.o crnn_layer.o coco_demo.o tag.o cifar.o yolo_demo.o go.o
+OBJ=gemm.o utils.o cuda.o deconvolutional_layer.o convolutional_layer.o list.o image.o activations.o im2col.o col2im.o blas.o crop_layer.o dropout_layer.o maxpool_layer.o softmax_layer.o data.o matrix.o network.o connected_layer.o cost_layer.o parser.o option_list.o darknet.o detection_layer.o imagenet.o captcha.o route_layer.o writing.o box.o nightmare.o normalization_layer.o avgpool_layer.o coco.o dice.o yolo.o layer.o compare.o classifier.o local_layer.o swag.o shortcut_layer.o activation_layer.o rnn_layer.o gru_layer.o rnn.o rnn_vid.o crnn_layer.o coco_demo.o tag.o cifar.o yolo_demo.o go.o batchnorm_layer.o
 ifeq ($(GPU), 1) 
+LDFLAGS+= -lstdc++ 
 OBJ+=convolutional_kernels.o deconvolutional_kernels.o activation_kernels.o im2col_kernels.o col2im_kernels.o blas_kernels.o crop_layer_kernels.o dropout_layer_kernels.o maxpool_layer_kernels.o softmax_layer_kernels.o network_kernels.o avgpool_layer_kernels.o
 endif
 

cfg/darknet.cfg

@@ -1,27 +1,20 @@
 [net]
 batch=128
 subdivisions=1
-height=256
-width=256
+height=224
+width=224
 channels=3
 momentum=0.9
 decay=0.0005
+max_crop=320
 
-learning_rate=0.01
-policy=sigmoid
-gamma=.00002
-step=400000
-max_batches=800000
-
-[crop]
-crop_height=224
-crop_width=224
-flip=1
-angle=0
-saturation=1
-exposure=1
+learning_rate=0.1
+policy=poly
+power=4
+max_batches=500000
 
 [convolutional]
+batch_normalize=1
 filters=16
 size=3
 stride=1
@@ -33,6 +26,7 @@ size=2
 stride=2
 
 [convolutional]
+batch_normalize=1
 filters=32
 size=3
 stride=1
@@ -44,6 +38,7 @@ size=2
 stride=2
 
 [convolutional]
+batch_normalize=1
 filters=64
 size=3
 stride=1
@@ -55,6 +50,7 @@ size=2
 stride=2
 
 [convolutional]
+batch_normalize=1
 filters=128
 size=3
 stride=1
@@ -66,6 +62,7 @@ size=2
 stride=2
 
 [convolutional]
+batch_normalize=1
 filters=256
 size=3
 stride=1
@@ -77,6 +74,7 @@ size=2
 stride=2
 
 [convolutional]
+batch_normalize=1
 filters=512
 size=3
 stride=1
@@ -88,6 +86,7 @@ size=2
 stride=2
 
 [convolutional]
+batch_normalize=1
 filters=1024
 size=3
 stride=1
@@ -96,9 +95,6 @@ activation=leaky
 
 [avgpool]
 
-[dropout]
-probability=.5
-
 [connected]
 output=1000
 activation=leaky

src/activation_kernels.cu

@@ -15,13 +15,19 @@ __device__ float elu_activate_kernel(float x){return (x >= 0)*x + (x < 0)*(exp(x
 __device__ float relie_activate_kernel(float x){return x*(x>0);}
 __device__ float ramp_activate_kernel(float x){return x*(x>0)+.1*x;}
 __device__ float leaky_activate_kernel(float x){return (x>0) ? x : .1*x;}
-__device__ float tanh_activate_kernel(float x){return (exp(2*x)-1)/(exp(2*x)+1);}
+__device__ float tanh_activate_kernel(float x){return (2/(1 + exp(-2*x)) - 1);}
 __device__ float plse_activate_kernel(float x)
 {
     if(x < -4) return .01 * (x + 4);
     if(x > 4)  return .01 * (x - 4) + 1;
     return .125*x + .5;
 }
+__device__ float stair_activate_kernel(float x)
+{
+    int n = floor(x);
+    if (n%2 == 0) return floor(x/2.);
+    else return (x - n) + floor(x/2.);
+}
 
 __device__ float linear_gradient_kernel(float x){return 1;}
 __device__ float logistic_gradient_kernel(float x){return (1-x)*x;}
@@ -37,6 +43,11 @@ __device__ float ramp_gradient_kernel(float x){return (x>0)+.1;}
 __device__ float leaky_gradient_kernel(float x){return (x>0) ? 1 : .1;}
 __device__ float tanh_gradient_kernel(float x){return 1-x*x;}
 __device__ float plse_gradient_kernel(float x){return (x < 0 || x > 1) ? .01 : .125;}
+__device__ float stair_gradient_kernel(float x)
+{
+    if (floor(x) == x) return 0;
+    return 1;
+}
 
 __device__ float activate_kernel(float x, ACTIVATION a)
 {
@@ -61,6 +72,8 @@ __device__ float activate_kernel(float x, ACTIVATION a)
             return tanh_activate_kernel(x);
         case PLSE:
             return plse_activate_kernel(x);
+        case STAIR:
+            return stair_activate_kernel(x);
     }
     return 0;
 }
@@ -88,6 +101,8 @@ __device__ float gradient_kernel(float x, ACTIVATION a)
             return tanh_gradient_kernel(x);
         case PLSE:
             return plse_gradient_kernel(x);
+        case STAIR:
+            return stair_gradient_kernel(x);
     }
     return 0;
 }

src/activations.c

@@ -28,6 +28,8 @@ char *get_activation_string(ACTIVATION a)
             return "plse";
         case LEAKY:
             return "leaky";
+        case STAIR:
+            return "stair";
         default:
             break;
     }
@@ -46,6 +48,7 @@ ACTIVATION get_activation(char *s)
     if (strcmp(s, "ramp")==0) return RAMP;
     if (strcmp(s, "leaky")==0) return LEAKY;
     if (strcmp(s, "tanh")==0) return TANH;
+    if (strcmp(s, "stair")==0) return STAIR;
     fprintf(stderr, "Couldn't find activation function %s, going with ReLU\n", s);
     return RELU;
 }
@@ -73,6 +76,8 @@ float activate(float x, ACTIVATION a)
             return tanh_activate(x);
         case PLSE:
             return plse_activate(x);
+        case STAIR:
+            return stair_activate(x);
     }
     return 0;
 }
@@ -108,6 +113,8 @@ float gradient(float x, ACTIVATION a)
             return tanh_gradient(x);
         case PLSE:
             return plse_gradient(x);
+        case STAIR:
+            return stair_gradient(x);
     }
     return 0;
 }

src/activations.h

@@ -4,7 +4,7 @@
 #include "math.h"
 
 typedef enum{
-    LOGISTIC, RELU, RELIE, LINEAR, RAMP, TANH, PLSE, LEAKY, ELU, LOGGY
+    LOGISTIC, RELU, RELIE, LINEAR, RAMP, TANH, PLSE, LEAKY, ELU, LOGGY, STAIR
 }ACTIVATION;
 
 ACTIVATION get_activation(char *s);
@@ -19,6 +19,12 @@ void activate_array_ongpu(float *x, int n, ACTIVATION a);
 void gradient_array_ongpu(float *x, int n, ACTIVATION a, float *delta);
 #endif
 
+static inline float stair_activate(float x)
+{
+    int n = floor(x);
+    if (n%2 == 0) return floor(x/2.);
+    else return (x - n) + floor(x/2.);
+}
 static inline float linear_activate(float x){return x;}
 static inline float logistic_activate(float x){return 1./(1. + exp(-x));}
 static inline float loggy_activate(float x){return 2./(1. + exp(-x)) - 1;}
@@ -42,6 +48,11 @@ static inline float loggy_gradient(float x)
     float y = (x+1.)/2.;
     return 2*(1-y)*y;
 }
+static inline float stair_gradient(float x)
+{
+    if (floor(x) == x) return 0;
+    return 1;
+}
 static inline float relu_gradient(float x){return (x>0);}
 static inline float elu_gradient(float x){return (x >= 0) + (x < 0)*(x + 1);}
 static inline float relie_gradient(float x){return (x>0) ? 1 : .01;}