diff --git a/Makefile b/Makefile
index 3bbaa5c0..b67ecacd 100644
--- a/Makefile
+++ b/Makefile
@@ -12,15 +12,18 @@ ARCH= -gencode arch=compute_20,code=[sm_20,sm_21] \
 # This is what I use, uncomment if you know your arch and want to specify
 ARCH=  -gencode arch=compute_52,code=compute_52
 
-VPATH=./src/
+VPATH=./src/:./examples
+LIB=libdarknet.a
 EXEC=darknet
 OBJDIR=./obj/
 
 CC=gcc
 NVCC=nvcc 
+AR=ar
+ARFLAGS=-rv
 OPTS=-Ofast
 LDFLAGS= -lm -pthread 
-COMMON= 
+COMMON= -Iinclude/
 CFLAGS=-Wall -Wfatal-errors 
 
 ifeq ($(DEBUG), 1) 
@@ -48,19 +51,25 @@ CFLAGS+= -DCUDNN
 LDFLAGS+= -lcudnn
 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 captcha.o route_layer.o writing.o box.o nightmare.o normalization_layer.o avgpool_layer.o coco.o dice.o yolo.o detector.o layer.o compare.o segmenter.o regressor.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 demo.o tag.o cifar.o go.o batchnorm_layer.o art.o region_layer.o reorg_layer.o lsd.o super.o voxel.o tree.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 detection_layer.o route_layer.o box.o normalization_layer.o avgpool_layer.o layer.o local_layer.o shortcut_layer.o activation_layer.o rnn_layer.o gru_layer.o crnn_layer.o demo.o batchnorm_layer.o region_layer.o reorg_layer.o tree.o 
+EXECOBJA=captcha.o lsd.o super.o voxel.o art.o tag.o cifar.o go.o rnn.o rnn_vid.o compare.o segmenter.o regressor.o classifier.o coco.o dice.o yolo.o detector.o  writing.o nightmare.o swag.o darknet.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 network_kernels.o avgpool_layer_kernels.o
 endif
 
+EXECOBJ = $(addprefix $(OBJDIR), $(EXECOBJA))
 OBJS = $(addprefix $(OBJDIR), $(OBJ))
-DEPS = $(wildcard src/*.h) Makefile
+DEPS = $(wildcard include/darknet/*.h) Makefile
 
-all: obj backup results $(EXEC)
+all: obj backup results $(LIB) $(EXEC)
 
-$(EXEC): $(OBJS)
-	$(CC) $(COMMON) $(CFLAGS) $^ -o $@ $(LDFLAGS)
+
+$(EXEC): $(EXECOBJ) $(LIB)
+	$(CC) $(COMMON) $(CFLAGS) $^ -o $@ $(LDFLAGS) $(LIB)
+
+$(LIB): $(OBJS)
+	$(AR) $(ARFLAGS) $@ $^ 
 
 $(OBJDIR)%.o: %.c $(DEPS)
 	$(CC) $(COMMON) $(CFLAGS) -c $< -o $@
@@ -78,5 +87,5 @@ results:
 .PHONY: clean
 
 clean:
-	rm -rf $(OBJS) $(EXEC)
+	rm -rf $(OBJS) $(LIB) $(EXEC) $(EXECOBJ)
 
diff --git a/cfg/imagenet1k.data b/cfg/imagenet1k.data
index 6634a274..b28c26c5 100644
--- a/cfg/imagenet1k.data
+++ b/cfg/imagenet1k.data
@@ -1,6 +1,6 @@
 classes=1000
 train  = /data/imagenet/imagenet1k.train.list
-valid  = /data/imagenet/imagenet1k.valid.list
+valid  = /data/imagenet/imagenet1k.train.list
 backup = /home/pjreddie/backup/
 labels = data/imagenet.labels.list
 names  = data/imagenet.shortnames.list
diff --git a/cfg/yolo.cfg b/cfg/yolo.cfg
index 2a0cd98f..088edf81 100644
--- a/cfg/yolo.cfg
+++ b/cfg/yolo.cfg
@@ -5,8 +5,8 @@ subdivisions=1
 # Training
 # batch=64
 # subdivisions=8
-width=416
-height=416
+width=608
+height=608
 channels=3
 momentum=0.9
 decay=0.0005
diff --git a/src/art.c b/examples/art.c
similarity index 89%
rename from src/art.c
rename to examples/art.c
index 694ae8e7..53fe60d0 100644
--- a/src/art.c
+++ b/examples/art.c
@@ -1,9 +1,9 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
-#include "option_list.h"
-#include "blas.h"
-#include "classifier.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/option_list.h"
+#include "darknet/blas.h"
+#include "darknet/classifier.h"
 #include <sys/time.h>
 
 void demo_art(char *cfgfile, char *weightfile, int cam_index)
diff --git a/src/captcha.c b/examples/captcha.c
similarity index 99%
rename from src/captcha.c
rename to examples/captcha.c
index 3d449b26..dce713ae 100644
--- a/src/captcha.c
+++ b/examples/captcha.c
@@ -1,6 +1,6 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
 
 void fix_data_captcha(data d, int mask)
 {
diff --git a/src/cifar.c b/examples/cifar.c
similarity index 98%
rename from src/cifar.c
rename to examples/cifar.c
index 5a6f96b9..fbbea285 100644
--- a/src/cifar.c
+++ b/examples/cifar.c
@@ -1,8 +1,8 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
-#include "option_list.h"
-#include "blas.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/option_list.h"
+#include "darknet/blas.h"
 
 void train_cifar(char *cfgfile, char *weightfile)
 {
diff --git a/src/classifier.c b/examples/classifier.c
similarity index 99%
rename from src/classifier.c
rename to examples/classifier.c
index 4f6a1a0b..440ab21a 100644
--- a/src/classifier.c
+++ b/examples/classifier.c
@@ -1,12 +1,12 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
-#include "option_list.h"
-#include "blas.h"
-#include "assert.h"
-#include "classifier.h"
-#include "cuda.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/option_list.h"
+#include "darknet/blas.h"
+#include "darknet/classifier.h"
+#include "darknet/cuda.h"
 #include <sys/time.h>
+#include <assert.h>
 
 float *get_regression_values(char **labels, int n)
 {
diff --git a/src/coco.c b/examples/coco.c
similarity index 98%
rename from src/coco.c
rename to examples/coco.c
index 3bcb6518..02e62455 100644
--- a/src/coco.c
+++ b/examples/coco.c
@@ -1,12 +1,12 @@
 #include <stdio.h>
 
-#include "network.h"
-#include "detection_layer.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
-#include "box.h"
-#include "demo.h"
+#include "darknet/network.h"
+#include "darknet/detection_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/box.h"
+#include "darknet/demo.h"
 
 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"};
 
diff --git a/src/darknet.c b/examples/darknet.c
similarity index 98%
rename from src/darknet.c
rename to examples/darknet.c
index b69cd436..4592dc3e 100644
--- a/src/darknet.c
+++ b/examples/darknet.c
@@ -2,11 +2,12 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#include "parser.h"
-#include "utils.h"
-#include "cuda.h"
-#include "blas.h"
-#include "connected_layer.h"
+#include "darknet/parser.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+#include "darknet/connected_layer.h"
+#include "darknet/convolutional_layer.h"
 
 extern void predict_classifier(char *datacfg, char *cfgfile, char *weightfile, char *filename, int top);
 extern void test_detector(char *datacfg, char *cfgfile, char *weightfile, char *filename, float thresh, float hier_thresh, char *outfile, int fullscreen);
@@ -172,7 +173,6 @@ void partial(char *cfgfile, char *weightfile, char *outfile, int max)
     save_weights_upto(net, outfile, max);
 }
 
-#include "convolutional_layer.h"
 void rescale_net(char *cfgfile, char *weightfile, char *outfile)
 {
     gpu_index = -1;
diff --git a/src/detector.c b/examples/detector.c
similarity index 98%
rename from src/detector.c
rename to examples/detector.c
index 12277b1a..2c8234d8 100644
--- a/src/detector.c
+++ b/examples/detector.c
@@ -1,12 +1,12 @@
-#include "network.h"
-#include "region_layer.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
-#include "box.h"
-#include "demo.h"
-#include "option_list.h"
-#include "blas.h"
+#include "darknet/network.h"
+#include "darknet/region_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/box.h"
+#include "darknet/demo.h"
+#include "darknet/option_list.h"
+#include "darknet/blas.h"
 
 static 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};
 
diff --git a/src/dice.c b/examples/dice.c
similarity index 97%
rename from src/dice.c
rename to examples/dice.c
index 22864598..7a2a1949 100644
--- a/src/dice.c
+++ b/examples/dice.c
@@ -1,6 +1,6 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
 
 char *dice_labels[] = {"face1","face2","face3","face4","face5","face6"};
 
diff --git a/src/go.c b/examples/go.c
similarity index 99%
rename from src/go.c
rename to examples/go.c
index c5c125cd..3e3d5f67 100644
--- a/src/go.c
+++ b/examples/go.c
@@ -1,9 +1,9 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
-#include "option_list.h"
-#include "blas.h"
-#include "data.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/option_list.h"
+#include "darknet/blas.h"
+#include "darknet/data.h"
 #include <unistd.h>
 
 int inverted = 1;
diff --git a/src/lsd.c b/examples/lsd.c
similarity index 99%
rename from src/lsd.c
rename to examples/lsd.c
index 312679ae..61f8e8f2 100644
--- a/src/lsd.c
+++ b/examples/lsd.c
@@ -1,8 +1,8 @@
-#include "network.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
-#include "blas.h"
+#include "darknet/network.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/blas.h"
 
 /*
 void train_lsd3(char *fcfg, char *fweight, char *gcfg, char *gweight, char *acfg, char *aweight, int clear)
diff --git a/src/nightmare.c b/examples/nightmare.c
similarity index 98%
rename from src/nightmare.c
rename to examples/nightmare.c
index 3ab735ce..3eda5fb6 100644
--- a/src/nightmare.c
+++ b/examples/nightmare.c
@@ -1,8 +1,8 @@
-#include "network.h"
-#include "parser.h"
-#include "blas.h"
-#include "utils.h"
-#include "region_layer.h"
+#include "darknet/network.h"
+#include "darknet/parser.h"
+#include "darknet/blas.h"
+#include "darknet/utils.h"
+#include "darknet/region_layer.h"
 
 // ./darknet nightmare cfg/extractor.recon.cfg ~/trained/yolo-coco.conv frame6.png -reconstruct -iters 500 -i 3 -lambda .1 -rate .01 -smooth 2
 
diff --git a/src/regressor.c b/examples/regressor.c
similarity index 97%
rename from src/regressor.c
rename to examples/regressor.c
index 93aeb982..30021e7f 100644
--- a/src/regressor.c
+++ b/examples/regressor.c
@@ -1,11 +1,11 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
-#include "option_list.h"
-#include "blas.h"
-#include "assert.h"
-#include "cuda.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/option_list.h"
+#include "darknet/blas.h"
+#include "darknet/cuda.h"
 #include <sys/time.h>
+#include <assert.h>
 
 void train_regressor(char *datacfg, char *cfgfile, char *weightfile, int *gpus, int ngpus, int clear)
 {
diff --git a/src/rnn.c b/examples/rnn.c
similarity index 98%
rename from src/rnn.c
rename to examples/rnn.c
index 62d7d8d1..ded23975 100644
--- a/src/rnn.c
+++ b/examples/rnn.c
@@ -1,8 +1,8 @@
-#include "network.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "blas.h"
-#include "parser.h"
+#include "darknet/network.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/blas.h"
+#include "darknet/parser.h"
 
 typedef struct {
     float *x;
@@ -152,7 +152,7 @@ void train_char_rnn(char *cfgfile, char *weightfile, char *filename, int clear,
     }
 
     int inputs = net.inputs;
-    fprintf(stderr, "Learning Rate: %g, Momentum: %g, Decay: %g\n", net.learning_rate, net.momentum, net.decay);
+    fprintf(stderr, "Learning Rate: %g, Momentum: %g, Decay: %g, Inputs: %d\n", net.learning_rate, net.momentum, net.decay, inputs);
     int batch = net.batch;
     int steps = net.time_steps;
     if(clear) *net.seen = 0;
diff --git a/src/rnn_vid.c b/examples/rnn_vid.c
similarity index 97%
rename from src/rnn_vid.c
rename to examples/rnn_vid.c
index 3042173f..a5e2a758 100644
--- a/src/rnn_vid.c
+++ b/examples/rnn_vid.c
@@ -1,8 +1,8 @@
-#include "network.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
-#include "blas.h"
+#include "darknet/network.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/blas.h"
 
 #ifdef OPENCV
 image get_image_from_stream(CvCapture *cap);
diff --git a/src/segmenter.c b/examples/segmenter.c
similarity index 97%
rename from src/segmenter.c
rename to examples/segmenter.c
index 32bb4843..12eb5408 100644
--- a/src/segmenter.c
+++ b/examples/segmenter.c
@@ -1,11 +1,11 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
-#include "option_list.h"
-#include "blas.h"
-#include "assert.h"
-#include "cuda.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/option_list.h"
+#include "darknet/blas.h"
+#include "darknet/cuda.h"
 #include <sys/time.h>
+#include <assert.h>
 
 void train_segmenter(char *datacfg, char *cfgfile, char *weightfile, int *gpus, int ngpus, int clear)
 {
diff --git a/src/super.c b/examples/super.c
similarity index 96%
rename from src/super.c
rename to examples/super.c
index b68fef57..d1f1bbf6 100644
--- a/src/super.c
+++ b/examples/super.c
@@ -1,7 +1,7 @@
-#include "network.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
+#include "darknet/network.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
 
 void train_super(char *cfgfile, char *weightfile, int clear)
 {
diff --git a/src/swag.c b/examples/swag.c
similarity index 93%
rename from src/swag.c
rename to examples/swag.c
index 74919693..91a7beb7 100644
--- a/src/swag.c
+++ b/examples/swag.c
@@ -1,9 +1,9 @@
-#include "network.h"
-#include "detection_layer.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
-#include "box.h"
+#include "darknet/network.h"
+#include "darknet/detection_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/box.h"
 
 void train_swag(char *cfgfile, char *weightfile)
 {
diff --git a/src/tag.c b/examples/tag.c
similarity index 98%
rename from src/tag.c
rename to examples/tag.c
index 084adba0..32a1bd14 100644
--- a/src/tag.c
+++ b/examples/tag.c
@@ -1,6 +1,6 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
 
 void train_tag(char *cfgfile, char *weightfile, int clear)
 {
diff --git a/src/voxel.c b/examples/voxel.c
similarity index 97%
rename from src/voxel.c
rename to examples/voxel.c
index af722e0f..b9387bc4 100644
--- a/src/voxel.c
+++ b/examples/voxel.c
@@ -1,7 +1,7 @@
-#include "network.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
+#include "darknet/network.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
 
 void extract_voxel(char *lfile, char *rfile, char *prefix)
 {
diff --git a/src/writing.c b/examples/writing.c
similarity index 98%
rename from src/writing.c
rename to examples/writing.c
index 789c0331..7136ff71 100644
--- a/src/writing.c
+++ b/examples/writing.c
@@ -1,6 +1,6 @@
-#include "network.h"
-#include "utils.h"
-#include "parser.h"
+#include "darknet/network.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
 
 void train_writing(char *cfgfile, char *weightfile)
 {
diff --git a/src/yolo.c b/examples/yolo.c
similarity index 98%
rename from src/yolo.c
rename to examples/yolo.c
index 6ddfb6ef..c1bef8bd 100644
--- a/src/yolo.c
+++ b/examples/yolo.c
@@ -1,10 +1,10 @@
-#include "network.h"
-#include "detection_layer.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
-#include "box.h"
-#include "demo.h"
+#include "darknet/network.h"
+#include "darknet/detection_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/box.h"
+#include "darknet/demo.h"
 
 char *voc_names[] = {"aeroplane", "bicycle", "bird", "boat", "bottle", "bus", "car", "cat", "chair", "cow", "diningtable", "dog", "horse", "motorbike", "person", "pottedplant", "sheep", "sofa", "train", "tvmonitor"};
 
diff --git a/src/activation_kernels.cu b/src/activation_kernels.cu
index 994e2068..f7882990 100644
--- a/src/activation_kernels.cu
+++ b/src/activation_kernels.cu
@@ -3,8 +3,8 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "activations.h"
-#include "cuda.h"
+#include "darknet/activations.h"
+#include "darknet/cuda.h"
 }
 
 
diff --git a/src/activation_layer.c b/src/activation_layer.c
index de4e4b7d..00596f8c 100644
--- a/src/activation_layer.c
+++ b/src/activation_layer.c
@@ -1,8 +1,8 @@
-#include "activation_layer.h"
-#include "utils.h"
-#include "cuda.h"
-#include "blas.h"
-#include "gemm.h"
+#include "darknet/activation_layer.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+#include "darknet/gemm.h"
 
 #include <math.h>
 #include <stdio.h>
diff --git a/src/activations.c b/src/activations.c
index 0cbb2f55..38a89f2d 100644
--- a/src/activations.c
+++ b/src/activations.c
@@ -1,4 +1,4 @@
-#include "activations.h"
+#include "darknet/activations.h"
 
 #include <math.h>
 #include <stdio.h>
diff --git a/src/avgpool_layer.c b/src/avgpool_layer.c
index 83034dbe..e7158e80 100644
--- a/src/avgpool_layer.c
+++ b/src/avgpool_layer.c
@@ -1,5 +1,5 @@
-#include "avgpool_layer.h"
-#include "cuda.h"
+#include "darknet/avgpool_layer.h"
+#include "darknet/cuda.h"
 #include <stdio.h>
 
 avgpool_layer make_avgpool_layer(int batch, int w, int h, int c)
diff --git a/src/avgpool_layer_kernels.cu b/src/avgpool_layer_kernels.cu
index a7eca3ae..d8f70adc 100644
--- a/src/avgpool_layer_kernels.cu
+++ b/src/avgpool_layer_kernels.cu
@@ -3,8 +3,8 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "avgpool_layer.h"
-#include "cuda.h"
+#include "darknet/avgpool_layer.h"
+#include "darknet/cuda.h"
 }
 
 __global__ void forward_avgpool_layer_kernel(int n, int w, int h, int c, float *input, float *output)
diff --git a/src/batchnorm_layer.c b/src/batchnorm_layer.c
index de081c4e..da8ac2fe 100644
--- a/src/batchnorm_layer.c
+++ b/src/batchnorm_layer.c
@@ -1,6 +1,6 @@
-#include "convolutional_layer.h"
-#include "batchnorm_layer.h"
-#include "blas.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/blas.h"
 #include <stdio.h>
 
 layer make_batchnorm_layer(int batch, int w, int h, int c)
@@ -135,10 +135,6 @@ void resize_batchnorm_layer(layer *layer, int w, int h)
 void forward_batchnorm_layer(layer l, network net)
 {
     if(l.type == BATCHNORM) copy_cpu(l.outputs*l.batch, net.input, 1, l.output, 1);
-    if(l.type == CONNECTED){
-        l.out_c = l.outputs;
-        l.out_h = l.out_w = 1;
-    }
     copy_cpu(l.outputs*l.batch, l.output, 1, l.x, 1);
     if(net.train){
         mean_cpu(l.output, l.batch, l.out_c, l.out_h*l.out_w, l.mean);
@@ -193,10 +189,6 @@ void push_batchnorm_layer(layer l)
 void forward_batchnorm_layer_gpu(layer l, network net)
 {
     if(l.type == BATCHNORM) copy_ongpu(l.outputs*l.batch, net.input_gpu, 1, l.output_gpu, 1);
-    if(l.type == CONNECTED){
-        l.out_c = l.outputs;
-        l.out_h = l.out_w = 1;
-    }
     copy_ongpu(l.outputs*l.batch, l.output_gpu, 1, l.x_gpu, 1);
     if (net.train) {
 #ifdef CUDNN
diff --git a/src/blas.c b/src/blas.c
index 1582f5fe..751c9d7a 100644
--- a/src/blas.c
+++ b/src/blas.c
@@ -1,5 +1,6 @@
-#include "blas.h"
-#include "math.h"
+#include "darknet/blas.h"
+
+#include <math.h>
 #include <assert.h>
 #include <float.h>
 #include <stdio.h>
diff --git a/src/blas_kernels.cu b/src/blas_kernels.cu
index 9f1337ca..b54603c9 100644
--- a/src/blas_kernels.cu
+++ b/src/blas_kernels.cu
@@ -4,9 +4,9 @@
 #include <assert.h>
 
 extern "C" {
-#include "blas.h"
-#include "cuda.h"
-#include "utils.h"
+#include "darknet/blas.h"
+#include "darknet/cuda.h"
+#include "darknet/utils.h"
 }
 
 __global__ void scale_bias_kernel(float *output, float *biases, int n, int size)
diff --git a/src/box.c b/src/box.c
index 88ca71ac..dbf432ee 100644
--- a/src/box.c
+++ b/src/box.c
@@ -1,4 +1,4 @@
-#include "box.h"
+#include "darknet/box.h"
 #include <stdio.h>
 #include <math.h>
 #include <stdlib.h>
diff --git a/src/col2im_kernels.cu b/src/col2im_kernels.cu
index aed2df9b..d8a5e371 100644
--- a/src/col2im_kernels.cu
+++ b/src/col2im_kernels.cu
@@ -3,8 +3,8 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "col2im.h"
-#include "cuda.h"
+#include "darknet/col2im.h"
+#include "darknet/cuda.h"
 }
 
 // src: https://github.com/BVLC/caffe/blob/master/src/caffe/util/im2col.cu
diff --git a/src/compare.c b/src/compare.c
index 4fd266c7..f773bfbb 100644
--- a/src/compare.c
+++ b/src/compare.c
@@ -1,11 +1,11 @@
 #include <stdio.h>
 
-#include "network.h"
-#include "detection_layer.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
-#include "box.h"
+#include "darknet/network.h"
+#include "darknet/detection_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/box.h"
 
 void train_compare(char *cfgfile, char *weightfile)
 {
diff --git a/src/connected_layer.c b/src/connected_layer.c
index f8c74235..4300dbdc 100644
--- a/src/connected_layer.c
+++ b/src/connected_layer.c
@@ -1,9 +1,9 @@
-#include "connected_layer.h"
-#include "batchnorm_layer.h"
-#include "utils.h"
-#include "cuda.h"
-#include "blas.h"
-#include "gemm.h"
+#include "darknet/connected_layer.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+#include "darknet/gemm.h"
 
 #include <math.h>
 #include <stdio.h>
diff --git a/src/convolutional_kernels.cu b/src/convolutional_kernels.cu
index b53dd16a..4146a3e5 100644
--- a/src/convolutional_kernels.cu
+++ b/src/convolutional_kernels.cu
@@ -3,14 +3,14 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "convolutional_layer.h"
-#include "batchnorm_layer.h"
-#include "gemm.h"
-#include "blas.h"
-#include "im2col.h"
-#include "col2im.h"
-#include "utils.h"
-#include "cuda.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/gemm.h"
+#include "darknet/blas.h"
+#include "darknet/im2col.h"
+#include "darknet/col2im.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
 }
 
 __global__ void binarize_kernel(float *x, int n, float *binary)
diff --git a/src/convolutional_layer.c b/src/convolutional_layer.c
index e5b5bb6f..0dbc7b74 100644
--- a/src/convolutional_layer.c
+++ b/src/convolutional_layer.c
@@ -1,10 +1,10 @@
-#include "convolutional_layer.h"
-#include "utils.h"
-#include "batchnorm_layer.h"
-#include "im2col.h"
-#include "col2im.h"
-#include "blas.h"
-#include "gemm.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/utils.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/im2col.h"
+#include "darknet/col2im.h"
+#include "darknet/blas.h"
+#include "darknet/gemm.h"
 #include <stdio.h>
 #include <time.h>
 
diff --git a/src/cost_layer.c b/src/cost_layer.c
index 76c001f0..251ec5a9 100644
--- a/src/cost_layer.c
+++ b/src/cost_layer.c
@@ -1,7 +1,7 @@
-#include "cost_layer.h"
-#include "utils.h"
-#include "cuda.h"
-#include "blas.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
 #include <math.h>
 #include <string.h>
 #include <stdlib.h>
diff --git a/src/crnn_layer.c b/src/crnn_layer.c
index 2478fef5..a8b040dd 100644
--- a/src/crnn_layer.c
+++ b/src/crnn_layer.c
@@ -1,9 +1,9 @@
-#include "crnn_layer.h"
-#include "convolutional_layer.h"
-#include "utils.h"
-#include "cuda.h"
-#include "blas.h"
-#include "gemm.h"
+#include "darknet/crnn_layer.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+#include "darknet/gemm.h"
 
 #include <math.h>
 #include <stdio.h>
diff --git a/src/crop_layer.c b/src/crop_layer.c
index 3b918529..afac7ee0 100644
--- a/src/crop_layer.c
+++ b/src/crop_layer.c
@@ -1,5 +1,5 @@
-#include "crop_layer.h"
-#include "cuda.h"
+#include "darknet/crop_layer.h"
+#include "darknet/cuda.h"
 #include <stdio.h>
 
 image get_crop_image(crop_layer l)
diff --git a/src/crop_layer_kernels.cu b/src/crop_layer_kernels.cu
index b6568219..41f6f456 100644
--- a/src/crop_layer_kernels.cu
+++ b/src/crop_layer_kernels.cu
@@ -3,10 +3,10 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "crop_layer.h"
-#include "utils.h"
-#include "cuda.h"
-#include "image.h"
+#include "darknet/crop_layer.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
+#include "darknet/image.h"
 }
 
 __device__ float get_pixel_kernel(float *image, int w, int h, int x, int y, int c)
diff --git a/src/cuda.c b/src/cuda.c
index 7e53d254..085a5764 100644
--- a/src/cuda.c
+++ b/src/cuda.c
@@ -2,10 +2,10 @@ int gpu_index = 0;
 
 #ifdef GPU
 
-#include "cuda.h"
-#include "utils.h"
-#include "blas.h"
-#include "assert.h"
+#include "darknet/cuda.h"
+#include "darknet/utils.h"
+#include "darknet/blas.h"
+#include <assert.h>
 #include <stdlib.h>
 #include <time.h>
 
diff --git a/src/data.c b/src/data.c
index dcc5089b..1ea115a3 100644
--- a/src/data.c
+++ b/src/data.c
@@ -1,7 +1,7 @@
-#include "data.h"
-#include "utils.h"
-#include "image.h"
-#include "cuda.h"
+#include "darknet/data.h"
+#include "darknet/utils.h"
+#include "darknet/image.h"
+#include "darknet/cuda.h"
 
 #include <stdio.h>
 #include <stdlib.h>
diff --git a/src/deconvolutional_kernels.cu b/src/deconvolutional_kernels.cu
index 16694634..1d4bc7c9 100644
--- a/src/deconvolutional_kernels.cu
+++ b/src/deconvolutional_kernels.cu
@@ -3,15 +3,15 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "convolutional_layer.h"
-#include "deconvolutional_layer.h"
-#include "batchnorm_layer.h"
-#include "gemm.h"
-#include "blas.h"
-#include "im2col.h"
-#include "col2im.h"
-#include "utils.h"
-#include "cuda.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/deconvolutional_layer.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/gemm.h"
+#include "darknet/blas.h"
+#include "darknet/im2col.h"
+#include "darknet/col2im.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
 }
 
 extern "C" void forward_deconvolutional_layer_gpu(layer l, network net)
diff --git a/src/deconvolutional_layer.c b/src/deconvolutional_layer.c
index 917d77fb..3ceff782 100644
--- a/src/deconvolutional_layer.c
+++ b/src/deconvolutional_layer.c
@@ -1,11 +1,12 @@
-#include "deconvolutional_layer.h"
-#include "convolutional_layer.h"
-#include "batchnorm_layer.h"
-#include "utils.h"
-#include "im2col.h"
-#include "col2im.h"
-#include "blas.h"
-#include "gemm.h"
+#include "darknet/deconvolutional_layer.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/utils.h"
+#include "darknet/im2col.h"
+#include "darknet/col2im.h"
+#include "darknet/blas.h"
+#include "darknet/gemm.h"
+
 #include <stdio.h>
 #include <time.h>
 
diff --git a/src/demo.c b/src/demo.c
index ff6c7329..cf22273b 100644
--- a/src/demo.c
+++ b/src/demo.c
@@ -1,12 +1,12 @@
-#include "network.h"
-#include "detection_layer.h"
-#include "region_layer.h"
-#include "cost_layer.h"
-#include "utils.h"
-#include "parser.h"
-#include "box.h"
-#include "image.h"
-#include "demo.h"
+#include "darknet/network.h"
+#include "darknet/detection_layer.h"
+#include "darknet/region_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
+#include "darknet/box.h"
+#include "darknet/image.h"
+#include "darknet/demo.h"
 #include <sys/time.h>
 
 #define DEMO 1
diff --git a/src/detection_layer.c b/src/detection_layer.c
index fd79cc70..721d64c2 100644
--- a/src/detection_layer.c
+++ b/src/detection_layer.c
@@ -1,10 +1,11 @@
-#include "detection_layer.h"
-#include "activations.h"
-#include "softmax_layer.h"
-#include "blas.h"
-#include "box.h"
-#include "cuda.h"
-#include "utils.h"
+#include "darknet/detection_layer.h"
+#include "darknet/activations.h"
+#include "darknet/softmax_layer.h"
+#include "darknet/blas.h"
+#include "darknet/box.h"
+#include "darknet/cuda.h"
+#include "darknet/utils.h"
+
 #include <stdio.h>
 #include <assert.h>
 #include <string.h>
diff --git a/src/dropout_layer.c b/src/dropout_layer.c
index 780554fb..1e344f45 100644
--- a/src/dropout_layer.c
+++ b/src/dropout_layer.c
@@ -1,6 +1,6 @@
-#include "dropout_layer.h"
-#include "utils.h"
-#include "cuda.h"
+#include "darknet/dropout_layer.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
 #include <stdlib.h>
 #include <stdio.h>
 
diff --git a/src/dropout_layer_kernels.cu b/src/dropout_layer_kernels.cu
index bd12b678..862e6f45 100644
--- a/src/dropout_layer_kernels.cu
+++ b/src/dropout_layer_kernels.cu
@@ -3,9 +3,9 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "dropout_layer.h"
-#include "cuda.h"
-#include "utils.h"
+#include "darknet/dropout_layer.h"
+#include "darknet/cuda.h"
+#include "darknet/utils.h"
 }
 
 __global__ void yoloswag420blazeit360noscope(float *input, int size, float *rand, float prob, float scale)
diff --git a/src/gemm.c b/src/gemm.c
index 3003be0e..245ae641 100644
--- a/src/gemm.c
+++ b/src/gemm.c
@@ -1,6 +1,6 @@
-#include "gemm.h"
-#include "utils.h"
-#include "cuda.h"
+#include "darknet/gemm.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
 #include <stdlib.h>
 #include <stdio.h>
 #include <math.h>
diff --git a/src/gru_layer.c b/src/gru_layer.c
index 7139f798..68ffed34 100644
--- a/src/gru_layer.c
+++ b/src/gru_layer.c
@@ -1,9 +1,9 @@
-#include "gru_layer.h"
-#include "connected_layer.h"
-#include "utils.h"
-#include "cuda.h"
-#include "blas.h"
-#include "gemm.h"
+#include "darknet/gru_layer.h"
+#include "darknet/connected_layer.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+#include "darknet/gemm.h"
 
 #include <math.h>
 #include <stdio.h>
diff --git a/src/im2col.c b/src/im2col.c
index 69ec98a9..67c078bc 100644
--- a/src/im2col.c
+++ b/src/im2col.c
@@ -1,4 +1,4 @@
-#include "im2col.h"
+#include "darknet/im2col.h"
 #include <stdio.h>
 float im2col_get_pixel(float *im, int height, int width, int channels,
                         int row, int col, int channel, int pad)
diff --git a/src/im2col_kernels.cu b/src/im2col_kernels.cu
index d42d600b..02f0e070 100644
--- a/src/im2col_kernels.cu
+++ b/src/im2col_kernels.cu
@@ -3,8 +3,8 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "im2col.h"
-#include "cuda.h"
+#include "darknet/im2col.h"
+#include "darknet/cuda.h"
 }
 
 // src: https://github.com/BVLC/caffe/blob/master/src/caffe/util/im2col.cu
diff --git a/src/image.c b/src/image.c
index 1a8c984a..b9f629ca 100644
--- a/src/image.c
+++ b/src/image.c
@@ -1,14 +1,14 @@
-#include "image.h"
-#include "utils.h"
-#include "blas.h"
-#include "cuda.h"
+#include "darknet/image.h"
+#include "darknet/utils.h"
+#include "darknet/blas.h"
+#include "darknet/cuda.h"
 #include <stdio.h>
 #include <math.h>
 
 #define STB_IMAGE_IMPLEMENTATION
-#include "stb_image.h"
+#include "darknet/stb_image.h"
 #define STB_IMAGE_WRITE_IMPLEMENTATION
-#include "stb_image_write.h"
+#include "darknet/stb_image_write.h"
 
 int windows = 0;
 
diff --git a/src/layer.c b/src/layer.c
index 791c5e77..bfa56e78 100644
--- a/src/layer.c
+++ b/src/layer.c
@@ -1,5 +1,6 @@
-#include "layer.h"
-#include "cuda.h"
+#include "darknet/layer.h"
+#include "darknet/cuda.h"
+
 #include <stdlib.h>
 
 void free_layer(layer l)
diff --git a/src/list.c b/src/list.c
index 0e4165d3..2be48d71 100644
--- a/src/list.c
+++ b/src/list.c
@@ -1,6 +1,6 @@
 #include <stdlib.h>
 #include <string.h>
-#include "list.h"
+#include "darknet/list.h"
 
 list *make_list()
 {
diff --git a/src/local_layer.c b/src/local_layer.c
index aad036e0..4fa934c5 100644
--- a/src/local_layer.c
+++ b/src/local_layer.c
@@ -1,9 +1,9 @@
-#include "local_layer.h"
-#include "utils.h"
-#include "im2col.h"
-#include "col2im.h"
-#include "blas.h"
-#include "gemm.h"
+#include "darknet/local_layer.h"
+#include "darknet/utils.h"
+#include "darknet/im2col.h"
+#include "darknet/col2im.h"
+#include "darknet/blas.h"
+#include "darknet/gemm.h"
 #include <stdio.h>
 #include <time.h>
 
diff --git a/src/matrix.c b/src/matrix.c
index 799916bf..6a0c4fd4 100644
--- a/src/matrix.c
+++ b/src/matrix.c
@@ -1,6 +1,6 @@
-#include "matrix.h"
-#include "utils.h"
-#include "blas.h"
+#include "darknet/matrix.h"
+#include "darknet/utils.h"
+#include "darknet/blas.h"
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
diff --git a/src/maxpool_layer.c b/src/maxpool_layer.c
index 17dedf7a..520dda54 100644
--- a/src/maxpool_layer.c
+++ b/src/maxpool_layer.c
@@ -1,5 +1,5 @@
-#include "maxpool_layer.h"
-#include "cuda.h"
+#include "darknet/maxpool_layer.h"
+#include "darknet/cuda.h"
 #include <stdio.h>
 
 image get_maxpool_image(maxpool_layer l)
diff --git a/src/maxpool_layer_kernels.cu b/src/maxpool_layer_kernels.cu
index 3202e84b..749c3880 100644
--- a/src/maxpool_layer_kernels.cu
+++ b/src/maxpool_layer_kernels.cu
@@ -3,8 +3,8 @@
 #include "cublas_v2.h"
 
 extern "C" {
-#include "maxpool_layer.h"
-#include "cuda.h"
+#include "darknet/maxpool_layer.h"
+#include "darknet/cuda.h"
 }
 
 __global__ void forward_maxpool_layer_kernel(int n, int in_h, int in_w, int in_c, int stride, int size, int pad, float *input, float *output, int *indexes)
diff --git a/src/network.c b/src/network.c
index 2b21338d..7b753549 100644
--- a/src/network.c
+++ b/src/network.c
@@ -1,34 +1,34 @@
 #include <stdio.h>
 #include <time.h>
 #include <assert.h>
-#include "network.h"
-#include "image.h"
-#include "data.h"
-#include "utils.h"
-#include "blas.h"
+#include "darknet/network.h"
+#include "darknet/image.h"
+#include "darknet/data.h"
+#include "darknet/utils.h"
+#include "darknet/blas.h"
 
-#include "crop_layer.h"
-#include "connected_layer.h"
-#include "gru_layer.h"
-#include "rnn_layer.h"
-#include "crnn_layer.h"
-#include "local_layer.h"
-#include "convolutional_layer.h"
-#include "activation_layer.h"
-#include "detection_layer.h"
-#include "region_layer.h"
-#include "normalization_layer.h"
-#include "batchnorm_layer.h"
-#include "maxpool_layer.h"
-#include "reorg_layer.h"
-#include "avgpool_layer.h"
-#include "cost_layer.h"
-#include "softmax_layer.h"
-#include "dropout_layer.h"
-#include "route_layer.h"
-#include "shortcut_layer.h"
-#include "parser.h"
-#include "data.h"
+#include "darknet/crop_layer.h"
+#include "darknet/connected_layer.h"
+#include "darknet/gru_layer.h"
+#include "darknet/rnn_layer.h"
+#include "darknet/crnn_layer.h"
+#include "darknet/local_layer.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/activation_layer.h"
+#include "darknet/detection_layer.h"
+#include "darknet/region_layer.h"
+#include "darknet/normalization_layer.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/maxpool_layer.h"
+#include "darknet/reorg_layer.h"
+#include "darknet/avgpool_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/softmax_layer.h"
+#include "darknet/dropout_layer.h"
+#include "darknet/route_layer.h"
+#include "darknet/shortcut_layer.h"
+#include "darknet/parser.h"
+#include "darknet/data.h"
 
 load_args get_base_args(network net)
 {
diff --git a/src/network_kernels.cu b/src/network_kernels.cu
index 2fbc2dd4..56ef50d7 100644
--- a/src/network_kernels.cu
+++ b/src/network_kernels.cu
@@ -7,32 +7,32 @@ extern "C" {
 #include <time.h>
 #include <assert.h>
 
-#include "network.h"
-#include "data.h"
-#include "utils.h"
-#include "parser.h"
+#include "darknet/network.h"
+#include "darknet/data.h"
+#include "darknet/utils.h"
+#include "darknet/parser.h"
 
-#include "crop_layer.h"
-#include "connected_layer.h"
-#include "rnn_layer.h"
-#include "gru_layer.h"
-#include "crnn_layer.h"
-#include "detection_layer.h"
-#include "region_layer.h"
-#include "convolutional_layer.h"
-#include "activation_layer.h"
-#include "maxpool_layer.h"
-#include "reorg_layer.h"
-#include "avgpool_layer.h"
-#include "normalization_layer.h"
-#include "batchnorm_layer.h"
-#include "cost_layer.h"
-#include "local_layer.h"
-#include "softmax_layer.h"
-#include "dropout_layer.h"
-#include "route_layer.h"
-#include "shortcut_layer.h"
-#include "blas.h"
+#include "darknet/crop_layer.h"
+#include "darknet/connected_layer.h"
+#include "darknet/rnn_layer.h"
+#include "darknet/gru_layer.h"
+#include "darknet/crnn_layer.h"
+#include "darknet/detection_layer.h"
+#include "darknet/region_layer.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/activation_layer.h"
+#include "darknet/maxpool_layer.h"
+#include "darknet/reorg_layer.h"
+#include "darknet/avgpool_layer.h"
+#include "darknet/normalization_layer.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/local_layer.h"
+#include "darknet/softmax_layer.h"
+#include "darknet/dropout_layer.h"
+#include "darknet/route_layer.h"
+#include "darknet/shortcut_layer.h"
+#include "darknet/blas.h"
 }
 
 void forward_network_gpu(network net)
diff --git a/src/normalization_layer.c b/src/normalization_layer.c
index 2be0d069..ce62e26e 100644
--- a/src/normalization_layer.c
+++ b/src/normalization_layer.c
@@ -1,5 +1,6 @@
-#include "normalization_layer.h"
-#include "blas.h"
+#include "darknet/normalization_layer.h"
+#include "darknet/blas.h"
+
 #include <stdio.h>
 
 layer make_normalization_layer(int batch, int w, int h, int c, int size, float alpha, float beta, float kappa)
diff --git a/src/option_list.c b/src/option_list.c
index f935af30..e3cae4d2 100644
--- a/src/option_list.c
+++ b/src/option_list.c
@@ -1,8 +1,8 @@
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
-#include "option_list.h"
-#include "utils.h"
+#include "darknet/option_list.h"
+#include "darknet/utils.h"
 
 list *read_data_cfg(char *filename)
 {
diff --git a/src/parser.c b/src/parser.c
index 47f2629a..38b1ccce 100644
--- a/src/parser.c
+++ b/src/parser.c
@@ -1,35 +1,35 @@
 #include <stdio.h>
 #include <string.h>
 #include <stdlib.h>
+#include <assert.h>
 
-#include "activation_layer.h"
-#include "activations.h"
-#include "assert.h"
-#include "avgpool_layer.h"
-#include "batchnorm_layer.h"
-#include "blas.h"
-#include "connected_layer.h"
-#include "deconvolutional_layer.h"
-#include "convolutional_layer.h"
-#include "cost_layer.h"
-#include "crnn_layer.h"
-#include "crop_layer.h"
-#include "detection_layer.h"
-#include "dropout_layer.h"
-#include "gru_layer.h"
-#include "list.h"
-#include "local_layer.h"
-#include "maxpool_layer.h"
-#include "normalization_layer.h"
-#include "option_list.h"
-#include "parser.h"
-#include "region_layer.h"
-#include "reorg_layer.h"
-#include "rnn_layer.h"
-#include "route_layer.h"
-#include "shortcut_layer.h"
-#include "softmax_layer.h"
-#include "utils.h"
+#include "darknet/activation_layer.h"
+#include "darknet/activations.h"
+#include "darknet/avgpool_layer.h"
+#include "darknet/batchnorm_layer.h"
+#include "darknet/blas.h"
+#include "darknet/connected_layer.h"
+#include "darknet/deconvolutional_layer.h"
+#include "darknet/convolutional_layer.h"
+#include "darknet/cost_layer.h"
+#include "darknet/crnn_layer.h"
+#include "darknet/crop_layer.h"
+#include "darknet/detection_layer.h"
+#include "darknet/dropout_layer.h"
+#include "darknet/gru_layer.h"
+#include "darknet/list.h"
+#include "darknet/local_layer.h"
+#include "darknet/maxpool_layer.h"
+#include "darknet/normalization_layer.h"
+#include "darknet/option_list.h"
+#include "darknet/parser.h"
+#include "darknet/region_layer.h"
+#include "darknet/reorg_layer.h"
+#include "darknet/rnn_layer.h"
+#include "darknet/route_layer.h"
+#include "darknet/shortcut_layer.h"
+#include "darknet/softmax_layer.h"
+#include "darknet/utils.h"
 
 typedef struct{
     char *type;
diff --git a/src/region_layer.c b/src/region_layer.c
index 9b645944..49147e35 100644
--- a/src/region_layer.c
+++ b/src/region_layer.c
@@ -1,9 +1,10 @@
-#include "region_layer.h"
-#include "activations.h"
-#include "blas.h"
-#include "box.h"
-#include "cuda.h"
-#include "utils.h"
+#include "darknet/region_layer.h"
+#include "darknet/activations.h"
+#include "darknet/blas.h"
+#include "darknet/box.h"
+#include "darknet/cuda.h"
+#include "darknet/utils.h"
+
 #include <stdio.h>
 #include <assert.h>
 #include <string.h>
diff --git a/src/reorg_layer.c b/src/reorg_layer.c
index 405a4266..d9f77154 100644
--- a/src/reorg_layer.c
+++ b/src/reorg_layer.c
@@ -1,6 +1,7 @@
-#include "reorg_layer.h"
-#include "cuda.h"
-#include "blas.h"
+#include "darknet/reorg_layer.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+
 #include <stdio.h>
 
 
diff --git a/src/rnn_layer.c b/src/rnn_layer.c
index fb4f1084..998bbacf 100644
--- a/src/rnn_layer.c
+++ b/src/rnn_layer.c
@@ -1,9 +1,9 @@
-#include "rnn_layer.h"
-#include "connected_layer.h"
-#include "utils.h"
-#include "cuda.h"
-#include "blas.h"
-#include "gemm.h"
+#include "darknet/rnn_layer.h"
+#include "darknet/connected_layer.h"
+#include "darknet/utils.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+#include "darknet/gemm.h"
 
 #include <math.h>
 #include <stdio.h>
diff --git a/src/route_layer.c b/src/route_layer.c
index 72267689..76f80f67 100644
--- a/src/route_layer.c
+++ b/src/route_layer.c
@@ -1,6 +1,7 @@
-#include "route_layer.h"
-#include "cuda.h"
-#include "blas.h"
+#include "darknet/route_layer.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+
 #include <stdio.h>
 
 route_layer make_route_layer(int batch, int n, int *input_layers, int *input_sizes)
diff --git a/src/shortcut_layer.c b/src/shortcut_layer.c
index 530d9980..bf60ddff 100644
--- a/src/shortcut_layer.c
+++ b/src/shortcut_layer.c
@@ -1,6 +1,7 @@
-#include "shortcut_layer.h"
-#include "cuda.h"
-#include "blas.h"
+#include "darknet/shortcut_layer.h"
+#include "darknet/cuda.h"
+#include "darknet/blas.h"
+
 #include <stdio.h>
 #include <assert.h>
 
diff --git a/src/softmax_layer.c b/src/softmax_layer.c
index 86b0237c..c0853f97 100644
--- a/src/softmax_layer.c
+++ b/src/softmax_layer.c
@@ -1,6 +1,7 @@
-#include "softmax_layer.h"
-#include "blas.h"
-#include "cuda.h"
+#include "darknet/softmax_layer.h"
+#include "darknet/blas.h"
+#include "darknet/cuda.h"
+
 #include <float.h>
 #include <math.h>
 #include <stdlib.h>
diff --git a/src/tree.c b/src/tree.c
index f36ca3a1..89b5a8b4 100644
--- a/src/tree.c
+++ b/src/tree.c
@@ -1,8 +1,8 @@
 #include <stdio.h>
 #include <stdlib.h>
-#include "tree.h"
-#include "utils.h"
-#include "data.h"
+#include "darknet/tree.h"
+#include "darknet/utils.h"
+#include "darknet/data.h"
 
 void change_leaves(tree *t, char *leaf_list)
 {
diff --git a/src/utils.c b/src/utils.c
index 9fa2d6bf..822fd331 100644
--- a/src/utils.c
+++ b/src/utils.c
@@ -7,7 +7,7 @@
 #include <float.h>
 #include <limits.h>
 
-#include "utils.h"
+#include "darknet/utils.h"
 
 int *read_intlist(char *gpu_list, int *ngpus, int d)
 {