network_predict

Closes #6

.gitignore

@@ -3,6 +3,8 @@ example/sample.jpg
 example/coco.names
 example/yolov3.cfg
 example/yolov3.weights
+example/yolov4.cfg
+example/yolov4.weights
 darknet.h
 *.so
 predictions.png

README.md

@@ -1,11 +1,13 @@
 [![GoDoc](https://godoc.org/github.com/LdDl/go-darknet?status.svg)](https://godoc.org/github.com/LdDl/go-darknet)
 
-# go-darknet: Go bindings for Darknet
+# go-darknet: Go bindings for Darknet (Yolo V4, Yolo V3)
-### go-darknet is a Go package, which uses Cgo to enable Go applications to use YOLO in [Darknet].
+### go-darknet is a Go package, which uses Cgo to enable Go applications to use YOLO V4/V3 in [Darknet].
 
 #### Since this repository https://github.com/gyonluks/go-darknet  is no longer maintained I decided to move on and make little different bindings for Darknet.
 #### This bindings aren't for [official implementation](https://github.com/pjreddie/darknet) but for [AlexeyAB's fork](https://github.com/AlexeyAB/darknet).
 
+#### Paper Yolo v4: https://arxiv.org/abs/2004.10934
+#### Paper Yolo v3: https://arxiv.org/abs/1804.02767
 
 ## Table of Contents
 
@@ -17,7 +19,7 @@
 
 ## Requirements
 
-For proper codebase please use fork of [darknet](https://github.com/AlexeyAB/darknet). Latest commit I've tested [here](https://github.com/AlexeyAB/darknet/commit/a234a5022333c930de08f2470184ef4e0c68356e)
+For proper codebase please use fork of [darknet](https://github.com/AlexeyAB/darknet). Latest commit I've tested [here](https://github.com/AlexeyAB/darknet/commit/9dc897d2c77d5ef43a6b237b717437375765b527)
 
 In order to use go-darknet, `libdarknet.so` should be available in one of
 the following locations:
@@ -55,27 +57,62 @@ Navigate to [example] folder
 cd $GOPATH/github.com/LdDl/go-darknet/example
 ```
 
-Download dataset (sample of image, coco.names, yolov3.cfg, yolov3.weights).
+Download dataset (sample of image, coco.names, yolov4.cfg (or v3), yolov4.weights(or v3)).
 ```shell
+#for yolo v4
 ./download_data.sh
+#for yolo v3
+./download_data_v3.sh
 ```
-Note: you don't need *coco.data* file anymore, because sh-script above does insert *coco.names* into 'names' filed in *yolov3.cfg* file (so AlexeyAB's fork can deal with it properly)
+Note: you don't need *coco.data* file anymore, because sh-script above does insert *coco.names* into 'names' filed in *yolov4.cfg* file (so AlexeyAB's fork can deal with it properly)
-So last rows in yolov3.cfg file will look like:
+So last rows in yolov4.cfg file will look like:
 ```bash
 ......
 [yolo]
-mask = 0,1,2
+.....
-anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+iou_loss=ciou
-classes=80
+nms_kind=greedynms
-num=9
+beta_nms=0.6
-jitter=.3
+
-ignore_thresh = .7
-truth_thresh = 1
-random=1
 names = coco.names # this is path to coco.names file
 ```
+Also do not forget change batch and subdivisions sizes from:
+```shell
+batch=64
+subdivisions=8
+```
+to
+```shell
+batch=1
+subdivisions=1
+```
+It will reduce amount of VRAM used for detector test.
+
 
 Build and run program
+Yolo V4:
+```shell
+go build main.go && ./main --configFile=yolov4.cfg --weightsFile=yolov4.weights --imageFile=sample.jpg
+```
+
+Output should be something like this:
+```shell
+traffic light (9): 73.5039% | start point: (238,73) | end point: (251, 106)
+truck (7): 96.6401% | start point: (95,79) | end point: (233, 287)
+truck (7): 96.4774% | start point: (662,158) | end point: (800, 321)
+truck (7): 96.1841% | start point: (0,77) | end point: (86, 333)
+truck (7): 46.8695% | start point: (434,173) | end point: (559, 216)
+car (2): 99.7370% | start point: (512,188) | end point: (741, 329)
+car (2): 99.2533% | start point: (260,191) | end point: (422, 322)
+car (2): 99.0333% | start point: (425,201) | end point: (547, 309)
+car (2): 83.3919% | start point: (386,210) | end point: (437, 287)
+car (2): 75.8621% | start point: (73,199) | end point: (102, 274)
+car (2): 39.1925% | start point: (386,206) | end point: (442, 240)
+bicycle (1): 76.3121% | start point: (189,298) | end point: (253, 402)
+person (0): 97.7213% | start point: (141,129) | end point: (283, 362)
+```
+
+Yolo V3:
 ```
 go build main.go && ./main --configFile=yolov3.cfg --weightsFile=yolov3.weights --imageFile=sample.jpg
 ```
@@ -96,6 +133,7 @@ car (2): 72.8053% | start point: (388,206) | end point: (437, 276)
 bicycle (1): 72.2932% | start point: (178,270) | end point: (268, 406)
 person (0): 97.3026% | start point: (143,135) | end point: (268, 343)
 ```
+
 ## Documentation
 
 See go-darknet's API documentation at [GoDoc].

docker/.env

@@ -0,0 +1 @@
+NAMESPACE=darknet

docker/Dockerfile

@@ -0,0 +1,40 @@
+# Build phase
+FROM ubuntu:18.04 as builder
+
+ENV darknet_commit=a234a5022333c930de08f2470184ef4e0c68356e
+
+WORKDIR /root/build
+COPY Makefile.cpu .
+RUN apt-get -y update && \
+	apt-get -y install --no-install-recommends git build-essential ca-certificates && \
+	git clone https://github.com/AlexeyAB/darknet && \
+	cd darknet && \
+	git checkout $darknet_commit && \
+	cp -f /root/build/Makefile.cpu Makefile && \
+	make
+
+# Final Image
+FROM golang:1.14
+
+RUN apt-get -y update && \
+        apt-get -y install --no-install-recommends nano bash jq
+
+WORKDIR /root
+COPY --from=builder /root/build/darknet/darknet \
+	/root/build/darknet/libdarknet.so \
+	/root/build/darknet/include/darknet.h \
+	./staging/
+
+RUN mv staging/darknet /usr/local/bin && \
+	mv staging/darknet.h /usr/include && \
+	mv staging/libdarknet.so /usr/lib && \
+	rm -rf staging
+
+RUN go get -u github.com/LdDl/go-darknet \
+    && go get -u github.com/disintegration/imaging
+
+WORKDIR /darknet
+
+COPY download_data.sh .
+
+CMD ["/bin/bash"]

docker/Dockerfile.gpu

@@ -0,0 +1,52 @@
+# Build phase
+FROM nvidia/cuda:10.0-cudnn7-devel-ubuntu18.04 as builder
+
+ENV darknet_commit=a234a5022333c930de08f2470184ef4e0c68356e
+
+WORKDIR /root/build
+COPY Makefile.gpu .
+RUN apt-get -y update && \
+	apt-get -y install git build-essential && \
+	git clone https://github.com/AlexeyAB/darknet.git && \
+	cd darknet && \
+	git checkout $darknet_commit && \
+        cp -f /root/build/Makefile.gpu Makefile && \
+	make
+
+# Final Image
+FROM nvidia/cuda:10.0-cudnn7-runtime-ubuntu18.04
+
+WORKDIR /root
+COPY --from=builder /root/build/darknet/darknet \
+	/root/build/darknet/libdarknet.so \
+	/root/build/darknet/include/darknet.h \
+	./staging/
+
+RUN mv staging/darknet /usr/local/bin && \
+	mv staging/darknet.h /usr/include && \
+	mv staging/libdarknet.so /usr/lib && \
+	rm -rf staging
+
+WORKDIR /tmp
+RUN cd /tmp \
+    && apt-get -y update \
+    && apt-get install -y wget git gcc \ 
+    && wget https://dl.google.com/go/go1.14.linux-amd64.tar.gz \
+    && tar -xvf go1.14.linux-amd64.tar.gz \
+    && mv go /usr/local
+
+RUN cp /usr/local/cuda-10.0/compat/* /usr/local/cuda-10.0/targets/x86_64-linux/lib/
+
+ENV GOROOT=/usr/local/go
+ENV GOPATH=/go
+ENV PATH=$GOPATH/bin:$GOROOT/bin:$PATH
+ENV LIBRARY_PATH=$LIBRARY_PATH:/usr/local/cuda-10.0/compat/
+
+RUN go get -u github.com/LdDl/go-darknet \
+    && go get -u github.com/disintegration/imaging
+
+WORKDIR /darknet
+
+COPY download_data.sh .
+
+CMD ["/bin/bash"]

docker/Makefile.cpu

@@ -0,0 +1,185 @@
+GPU=0
+CUDNN=0
+CUDNN_HALF=0
+OPENCV=0
+AVX=1
+OPENMP=1
+LIBSO=1
+ZED_CAMERA=0 # ZED SDK 3.0 and above
+ZED_CAMERA_v2_8=0 # ZED SDK 2.X
+
+# set GPU=1 and CUDNN=1 to speedup on GPU
+# set CUDNN_HALF=1 to further speedup 3 x times (Mixed-precision on Tensor Cores) GPU: Volta, Xavier, Turing and higher
+# set AVX=1 and OPENMP=1 to speedup on CPU (if error occurs then set AVX=0)
+
+USE_CPP=0
+DEBUG=0
+
+ARCH= -gencode arch=compute_30,code=sm_30 \
+      -gencode arch=compute_35,code=sm_35 \
+      -gencode arch=compute_50,code=[sm_50,compute_50] \
+      -gencode arch=compute_52,code=[sm_52,compute_52] \
+	  -gencode arch=compute_61,code=[sm_61,compute_61]
+
+OS := $(shell uname)
+
+# Tesla V100
+# ARCH= -gencode arch=compute_70,code=[sm_70,compute_70]
+
+# GeForce RTX 2080 Ti, RTX 2080, RTX 2070, Quadro RTX 8000, Quadro RTX 6000, Quadro RTX 5000, Tesla T4, XNOR Tensor Cores
+# ARCH= -gencode arch=compute_75,code=[sm_75,compute_75]
+
+# Jetson XAVIER
+# ARCH= -gencode arch=compute_72,code=[sm_72,compute_72]
+
+# GTX 1080, GTX 1070, GTX 1060, GTX 1050, GTX 1030, Titan Xp, Tesla P40, Tesla P4
+# ARCH= -gencode arch=compute_61,code=sm_61 -gencode arch=compute_61,code=compute_61
+
+# GP100/Tesla P100 - DGX-1
+# ARCH= -gencode arch=compute_60,code=sm_60
+
+# For Jetson TX1, Tegra X1, DRIVE CX, DRIVE PX - uncomment:
+# ARCH= -gencode arch=compute_53,code=[sm_53,compute_53]
+
+# For Jetson Tx2 or Drive-PX2 uncomment:
+# ARCH= -gencode arch=compute_62,code=[sm_62,compute_62]
+
+
+# VPATH=./src/
+VPATH=./src/:./examples
+SLIB=libdarknet.so
+
+EXEC=darknet
+OBJDIR=./obj/
+
+ifeq ($(LIBSO), 1)
+LIBNAMESO=libdarknet.so
+APPNAMESO=uselib
+endif
+
+ifeq ($(USE_CPP), 1)
+CC=g++
+else
+CC=gcc
+endif
+
+CPP=g++ -std=c++11
+NVCC=nvcc
+OPTS=-Ofast
+LDFLAGS= -lm -pthread
+COMMON= -Iinclude/ -I3rdparty/stb/include
+CFLAGS=-Wall -Wfatal-errors -Wno-unused-result -Wno-unknown-pragmas -fPIC
+
+ifeq ($(DEBUG), 1)
+#OPTS= -O0 -g
+#OPTS= -Og -g
+COMMON+= -DDEBUG
+CFLAGS+= -DDEBUG
+else
+ifeq ($(AVX), 1)
+CFLAGS+= -ffp-contract=fast -mavx -mavx2 -msse3 -msse4.1 -msse4.2 -msse4a
+endif
+endif
+
+CFLAGS+=$(OPTS)
+
+ifneq (,$(findstring MSYS_NT,$(OS)))
+LDFLAGS+=-lws2_32
+endif
+
+ifeq ($(OPENCV), 1)
+COMMON+= -DOPENCV
+CFLAGS+= -DOPENCV
+LDFLAGS+= `pkg-config --libs opencv4 2> /dev/null || pkg-config --libs opencv`
+COMMON+= `pkg-config --cflags opencv4 2> /dev/null || pkg-config --cflags opencv`
+endif
+
+ifeq ($(OPENMP), 1)
+CFLAGS+= -fopenmp
+LDFLAGS+= -lgomp
+endif
+
+ifeq ($(GPU), 1)
+COMMON+= -DGPU -I/usr/local/cuda/include/
+CFLAGS+= -DGPU
+ifeq ($(OS),Darwin) #MAC
+LDFLAGS+= -L/usr/local/cuda/lib -lcuda -lcudart -lcublas -lcurand
+else
+LDFLAGS+= -L/usr/local/cuda/lib64 -lcuda -lcudart -lcublas -lcurand
+endif
+endif
+
+ifeq ($(CUDNN), 1)
+COMMON+= -DCUDNN
+ifeq ($(OS),Darwin) #MAC
+CFLAGS+= -DCUDNN -I/usr/local/cuda/include
+LDFLAGS+= -L/usr/local/cuda/lib -lcudnn
+else
+CFLAGS+= -DCUDNN -I/usr/local/cudnn/include
+LDFLAGS+= -L/usr/local/cudnn/lib64 -lcudnn
+endif
+endif
+
+ifeq ($(CUDNN_HALF), 1)
+COMMON+= -DCUDNN_HALF
+CFLAGS+= -DCUDNN_HALF
+ARCH+= -gencode arch=compute_70,code=[sm_70,compute_70]
+endif
+
+ifeq ($(ZED_CAMERA), 1)
+CFLAGS+= -DZED_STEREO -I/usr/local/zed/include
+ifeq ($(ZED_CAMERA_v2_8), 1)
+LDFLAGS+= -L/usr/local/zed/lib -lsl_core -lsl_input -lsl_zed
+#-lstdc++ -D_GLIBCXX_USE_CXX11_ABI=0 
+else
+LDFLAGS+= -L/usr/local/zed/lib -lsl_zed
+#-lstdc++ -D_GLIBCXX_USE_CXX11_ABI=0 
+endif
+endif
+
+OBJ=image_opencv.o http_stream.o gemm.o utils.o dark_cuda.o convolutional_layer.o list.o image.o activations.o im2col.o col2im.o blas.o crop_layer.o dropout_layer.o maxpool_layer.o softmax_layer.o data.o matrix.o network.o connected_layer.o cost_layer.o parser.o option_list.o darknet.o detection_layer.o captcha.o route_layer.o writing.o box.o nightmare.o normalization_layer.o avgpool_layer.o coco.o dice.o yolo.o detector.o layer.o compare.o classifier.o local_layer.o swag.o shortcut_layer.o activation_layer.o rnn_layer.o gru_layer.o rnn.o rnn_vid.o crnn_layer.o demo.o tag.o cifar.o go.o batchnorm_layer.o art.o region_layer.o reorg_layer.o reorg_old_layer.o super.o voxel.o tree.o yolo_layer.o gaussian_yolo_layer.o upsample_layer.o lstm_layer.o conv_lstm_layer.o scale_channels_layer.o sam_layer.o
+ifeq ($(GPU), 1) 
+LDFLAGS+= -lstdc++ 
+OBJ+=convolutional_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
+
+OBJS = $(addprefix $(OBJDIR), $(OBJ))
+DEPS = $(wildcard src/*.h) Makefile include/darknet.h
+
+all: $(OBJDIR) backup results setchmod $(EXEC) $(LIBNAMESO) $(APPNAMESO)
+
+ifeq ($(LIBSO), 1)
+CFLAGS+= -fPIC
+
+$(LIBNAMESO): $(OBJDIR) $(OBJS) include/yolo_v2_class.hpp src/yolo_v2_class.cpp
+	$(CPP) -shared -std=c++11 -fvisibility=hidden -DLIB_EXPORTS $(COMMON) $(CFLAGS) $(OBJS) src/yolo_v2_class.cpp -o $@ $(LDFLAGS)
+
+$(APPNAMESO): $(LIBNAMESO) include/yolo_v2_class.hpp src/yolo_console_dll.cpp
+	$(CPP) -std=c++11 $(COMMON) $(CFLAGS) -o $@ src/yolo_console_dll.cpp $(LDFLAGS) -L ./ -l:$(LIBNAMESO)
+endif
+
+$(EXEC): $(OBJS)
+	$(CPP) -std=c++11 $(COMMON) $(CFLAGS) $^ -o $@ $(LDFLAGS)
+
+$(OBJDIR)%.o: %.c $(DEPS)
+	$(CC) $(COMMON) $(CFLAGS) -c $< -o $@
+
+$(OBJDIR)%.o: %.cpp $(DEPS)
+	$(CPP) -std=c++11 $(COMMON) $(CFLAGS) -c $< -o $@
+
+$(OBJDIR)%.o: %.cu $(DEPS)
+	$(NVCC) $(ARCH) $(COMMON) --compiler-options "$(CFLAGS)" -c $< -o $@
+
+$(OBJDIR):
+	mkdir -p $(OBJDIR)
+backup:
+	mkdir -p backup
+results:
+	mkdir -p results
+setchmod:
+	chmod +x *.sh
+
+.PHONY: clean
+
+clean:
+	rm -rf $(OBJS) $(EXEC) $(LIBNAMESO) $(APPNAMESO)

docker/Makefile.gpu

@@ -0,0 +1,186 @@
+
+GPU=1
+CUDNN=1
+CUDNN_HALF=0
+OPENCV=0
+AVX=0
+OPENMP=0
+LIBSO=1
+ZED_CAMERA=0 # ZED SDK 3.0 and above
+ZED_CAMERA_v2_8=0 # ZED SDK 2.X
+
+# set GPU=1 and CUDNN=1 to speedup on GPU
+# set CUDNN_HALF=1 to further speedup 3 x times (Mixed-precision on Tensor Cores) GPU: Volta, Xavier, Turing and higher
+# set AVX=1 and OPENMP=1 to speedup on CPU (if error occurs then set AVX=0)
+
+USE_CPP=0
+DEBUG=0
+
+ARCH= -gencode arch=compute_30,code=sm_30 \
+      -gencode arch=compute_35,code=sm_35 \
+      -gencode arch=compute_50,code=[sm_50,compute_50] \
+      -gencode arch=compute_52,code=[sm_52,compute_52] \
+	  -gencode arch=compute_61,code=[sm_61,compute_61]
+
+OS := $(shell uname)
+
+# Tesla V100
+# ARCH= -gencode arch=compute_70,code=[sm_70,compute_70]
+
+# GeForce RTX 2080 Ti, RTX 2080, RTX 2070, Quadro RTX 8000, Quadro RTX 6000, Quadro RTX 5000, Tesla T4, XNOR Tensor Cores
+# ARCH= -gencode arch=compute_75,code=[sm_75,compute_75]
+
+# Jetson XAVIER
+# ARCH= -gencode arch=compute_72,code=[sm_72,compute_72]
+
+# GTX 1080, GTX 1070, GTX 1060, GTX 1050, GTX 1030, Titan Xp, Tesla P40, Tesla P4
+# ARCH= -gencode arch=compute_61,code=sm_61 -gencode arch=compute_61,code=compute_61
+
+# GP100/Tesla P100 - DGX-1
+# ARCH= -gencode arch=compute_60,code=sm_60
+
+# For Jetson TX1, Tegra X1, DRIVE CX, DRIVE PX - uncomment:
+# ARCH= -gencode arch=compute_53,code=[sm_53,compute_53]
+
+# For Jetson Tx2 or Drive-PX2 uncomment:
+# ARCH= -gencode arch=compute_62,code=[sm_62,compute_62]
+
+
+# VPATH=./src/
+VPATH=./src/:./examples
+SLIB=libdarknet.so
+
+EXEC=darknet
+OBJDIR=./obj/
+
+ifeq ($(LIBSO), 1)
+LIBNAMESO=libdarknet.so
+APPNAMESO=uselib
+endif
+
+ifeq ($(USE_CPP), 1)
+CC=g++
+else
+CC=gcc
+endif
+
+CPP=g++ -std=c++11
+NVCC=nvcc
+OPTS=-Ofast
+LDFLAGS= -lm -pthread
+COMMON= -Iinclude/ -I3rdparty/stb/include
+CFLAGS=-Wall -Wfatal-errors -Wno-unused-result -Wno-unknown-pragmas -fPIC
+
+ifeq ($(DEBUG), 1)
+#OPTS= -O0 -g
+#OPTS= -Og -g
+COMMON+= -DDEBUG
+CFLAGS+= -DDEBUG
+else
+ifeq ($(AVX), 1)
+CFLAGS+= -ffp-contract=fast -mavx -mavx2 -msse3 -msse4.1 -msse4.2 -msse4a
+endif
+endif
+
+CFLAGS+=$(OPTS)
+
+ifneq (,$(findstring MSYS_NT,$(OS)))
+LDFLAGS+=-lws2_32
+endif
+
+ifeq ($(OPENCV), 1)
+COMMON+= -DOPENCV
+CFLAGS+= -DOPENCV
+LDFLAGS+= `pkg-config --libs opencv4 2> /dev/null || pkg-config --libs opencv`
+COMMON+= `pkg-config --cflags opencv4 2> /dev/null || pkg-config --cflags opencv`
+endif
+
+ifeq ($(OPENMP), 1)
+CFLAGS+= -fopenmp
+LDFLAGS+= -lgomp
+endif
+
+ifeq ($(GPU), 1)
+COMMON+= -DGPU -I/usr/local/cuda/include/
+CFLAGS+= -DGPU
+ifeq ($(OS),Darwin) #MAC
+LDFLAGS+= -L/usr/local/cuda/lib -lcuda -lcudart -lcublas -lcurand
+else
+LDFLAGS+= -L/usr/local/cuda/lib64 -lcuda -lcudart -lcublas -lcurand
+endif
+endif
+
+ifeq ($(CUDNN), 1)
+COMMON+= -DCUDNN
+ifeq ($(OS),Darwin) #MAC
+CFLAGS+= -DCUDNN -I/usr/local/cuda/include
+LDFLAGS+= -L/usr/local/cuda/lib -lcudnn
+else
+CFLAGS+= -DCUDNN -I/usr/local/cudnn/include
+LDFLAGS+= -L/usr/local/cudnn/lib64 -lcudnn
+endif
+endif
+
+ifeq ($(CUDNN_HALF), 1)
+COMMON+= -DCUDNN_HALF
+CFLAGS+= -DCUDNN_HALF
+ARCH+= -gencode arch=compute_70,code=[sm_70,compute_70]
+endif
+
+ifeq ($(ZED_CAMERA), 1)
+CFLAGS+= -DZED_STEREO -I/usr/local/zed/include
+ifeq ($(ZED_CAMERA_v2_8), 1)
+LDFLAGS+= -L/usr/local/zed/lib -lsl_core -lsl_input -lsl_zed
+#-lstdc++ -D_GLIBCXX_USE_CXX11_ABI=0 
+else
+LDFLAGS+= -L/usr/local/zed/lib -lsl_zed
+#-lstdc++ -D_GLIBCXX_USE_CXX11_ABI=0 
+endif
+endif
+
+OBJ=image_opencv.o http_stream.o gemm.o utils.o dark_cuda.o convolutional_layer.o list.o image.o activations.o im2col.o col2im.o blas.o crop_layer.o dropout_layer.o maxpool_layer.o softmax_layer.o data.o matrix.o network.o connected_layer.o cost_layer.o parser.o option_list.o darknet.o detection_layer.o captcha.o route_layer.o writing.o box.o nightmare.o normalization_layer.o avgpool_layer.o coco.o dice.o yolo.o detector.o layer.o compare.o classifier.o local_layer.o swag.o shortcut_layer.o activation_layer.o rnn_layer.o gru_layer.o rnn.o rnn_vid.o crnn_layer.o demo.o tag.o cifar.o go.o batchnorm_layer.o art.o region_layer.o reorg_layer.o reorg_old_layer.o super.o voxel.o tree.o yolo_layer.o gaussian_yolo_layer.o upsample_layer.o lstm_layer.o conv_lstm_layer.o scale_channels_layer.o sam_layer.o
+ifeq ($(GPU), 1) 
+LDFLAGS+= -lstdc++ 
+OBJ+=convolutional_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
+
+OBJS = $(addprefix $(OBJDIR), $(OBJ))
+DEPS = $(wildcard src/*.h) Makefile include/darknet.h
+
+all: $(OBJDIR) backup results setchmod $(EXEC) $(LIBNAMESO) $(APPNAMESO)
+
+ifeq ($(LIBSO), 1)
+CFLAGS+= -fPIC
+
+$(LIBNAMESO): $(OBJDIR) $(OBJS) include/yolo_v2_class.hpp src/yolo_v2_class.cpp
+	$(CPP) -shared -std=c++11 -fvisibility=hidden -DLIB_EXPORTS $(COMMON) $(CFLAGS) $(OBJS) src/yolo_v2_class.cpp -o $@ $(LDFLAGS)
+
+$(APPNAMESO): $(LIBNAMESO) include/yolo_v2_class.hpp src/yolo_console_dll.cpp
+	$(CPP) -std=c++11 $(COMMON) $(CFLAGS) -o $@ src/yolo_console_dll.cpp $(LDFLAGS) -L ./ -l:$(LIBNAMESO)
+endif
+
+$(EXEC): $(OBJS)
+	$(CPP) -std=c++11 $(COMMON) $(CFLAGS) $^ -o $@ $(LDFLAGS)
+
+$(OBJDIR)%.o: %.c $(DEPS)
+	$(CC) $(COMMON) $(CFLAGS) -c $< -o $@
+
+$(OBJDIR)%.o: %.cpp $(DEPS)
+	$(CPP) -std=c++11 $(COMMON) $(CFLAGS) -c $< -o $@
+
+$(OBJDIR)%.o: %.cu $(DEPS)
+	$(NVCC) $(ARCH) $(COMMON) --compiler-options "$(CFLAGS)" -c $< -o $@
+
+$(OBJDIR):
+	mkdir -p $(OBJDIR)
+backup:
+	mkdir -p backup
+results:
+	mkdir -p results
+setchmod:
+	chmod +x *.sh
+
+.PHONY: clean
+
+clean:
+	rm -rf $(OBJS) $(EXEC) $(LIBNAMESO) $(APPNAMESO)

docker/docker-compose.yml

@@ -0,0 +1,32 @@
+---
+version: '3.7'
+services:
+
+  sidekiq: &darknet_base 
+    container_name: ${NAMESPACE}-sidekiq
+    build:
+      context: .
+      dockerfile: Dockerfile
+    image: go-darknet:latest
+    working_dir: /darknet
+    volumes:
+    - darknet-data:/darknet/models
+    command: /darknet/download_data.sh
+
+  darknet:
+    <<: *darknet_base
+    container_name: ${NAMESPACE}-api
+    ports:
+    - "9003:9003"
+    restart: unless-stopped
+    depends_on:
+    - sidekiq
+    command: ["/bin/bash"]
+    # command: ["darknet-server"]
+
+volumes:
+  darknet-data:
+    driver_opts:
+      type: none
+      o: bind
+      device: ${PWD}/models

docker/download_data.sh

@@ -0,0 +1,10 @@
+#!/bin/sh
+
+# set -x
+# set -e
+
+wget -nc --output-document=sample.jpg https://cdn-images-1.medium.com/max/800/1*EYFejGUjvjPcc4PZTwoufw.jpeg
+wget -nc --output-document=./models/coco.names https://raw.githubusercontent.com/AlexeyAB/darknet/master/data/coco.names
+wget -nc --output-document=./models/yolov3.cfg https://raw.githubusercontent.com/AlexeyAB/darknet/master/cfg/yolov3.cfg
+sed -i -e "\$anames = coco.names" ./models/yolov3.cfg
+wget -nc --output-document=./models/yolov3.weights https://pjreddie.com/media/files/yolov3.weights

docker/models/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore

example/download_data.sh

@@ -1,5 +1,5 @@
 wget --output-document=sample.jpg https://cdn-images-1.medium.com/max/800/1*EYFejGUjvjPcc4PZTwoufw.jpeg
 wget --output-document=coco.names https://raw.githubusercontent.com/AlexeyAB/darknet/master/data/coco.names
-wget --output-document=yolov3.cfg https://raw.githubusercontent.com/AlexeyAB/darknet/master/cfg/yolov3.cfg
+wget --output-document=yolov4.cfg https://raw.githubusercontent.com/AlexeyAB/darknet/master/cfg/yolov4.cfg
-sed -i -e "\$anames = coco.names" yolov3.cfg
+sed -i -e "\$anames = coco.names" yolov4.cfg
-wget --output-document=yolov3.weights https://pjreddie.com/media/files/yolov3.weights
+wget --output-document=yolov4.weights https://github.com/AlexeyAB/darknet/releases/download/darknet_yolo_v3_optimal/yolov4.weights

example/download_data_v3.sh

@@ -0,0 +1,5 @@
+wget --output-document=sample.jpg https://cdn-images-1.medium.com/max/800/1*EYFejGUjvjPcc4PZTwoufw.jpeg
+wget --output-document=coco.names https://raw.githubusercontent.com/AlexeyAB/darknet/master/data/coco.names
+wget --output-document=yolov3.cfg https://raw.githubusercontent.com/AlexeyAB/darknet/master/cfg/yolov3.cfg
+sed -i -e "\$anames = coco.names" yolov3.cfg
+wget --output-document=yolov3.weights https://pjreddie.com/media/files/yolov3.weights

image.go

@@ -14,12 +14,14 @@ import (
 type DarknetImage struct {
 	Width  int
 	Height int
+	ans    []float32
 	image  C.image
 }
 
 // Close and release resources.
 func (img *DarknetImage) Close() error {
 	C.free_image(img.image)
+	img.ans = nil
 	return nil
 }
 
@@ -29,10 +31,10 @@ func imgTofloat32(src image.Image) []float32 {
 	width, height := bounds.Max.X, bounds.Max.Y
 	srcRGBA := image.NewRGBA(src.Bounds())
 	draw.Copy(srcRGBA, image.Point{}, src, src.Bounds(), draw.Src, nil)
-	ans := []float32{}
+
-	red := []float32{}
+	red := make([]float32, 0, width*height)
-	green := []float32{}
+	green := make([]float32, 0, width*height)
-	blue := []float32{}
+	blue := make([]float32, 0, width*height)
 	for y := 0; y < height; y++ {
 		for x := 0; x < width; x++ {
 			idxSource := (y*width + x) * 4
@@ -43,22 +45,40 @@ func imgTofloat32(src image.Image) []float32 {
 			blue = append(blue, bpix)
 		}
 	}
-	ans = append(ans, red...)
+	srcRGBA = nil
-	ans = append(ans, green...)
+
-	ans = append(ans, blue...)
+	ans := make([]float32, len(red)+len(green)+len(blue))
+	copy(ans[:len(red)], red)
+	copy(ans[len(red):len(red)+len(green)], green)
+	copy(ans[len(red)+len(green):], blue)
+	red = nil
+	green = nil
+	blue = nil
 	return ans
 }
 
 // Image2Float32 Returns []float32 representation of image.Image
 func Image2Float32(img image.Image) (*DarknetImage, error) {
-	ans := imgTofloat32(img)
+	// ans := imgTofloat32(img)
 	width := img.Bounds().Dx()
 	height := img.Bounds().Dy()
 	imgDarknet := &DarknetImage{
 		Width:  width,
 		Height: height,
+		ans:    imgTofloat32(img),
 		image:  C.make_image(C.int(width), C.int(height), 3),
 	}
-	C.fill_image_f32(&imgDarknet.image, C.int(width), C.int(height), 3, (*C.float)(unsafe.Pointer(&ans[0])))
+	C.fill_image_f32(&imgDarknet.image, C.int(width), C.int(height), 3, (*C.float)(unsafe.Pointer(&imgDarknet.ans[0])))
+	return imgDarknet, nil
+}
+
+// Float32ToDarknetImage Converts []float32 to darknet image
+func Float32ToDarknetImage(flatten []float32, width, height int) (*DarknetImage, error) {
+	imgDarknet := &DarknetImage{
+		Width:  width,
+		Height: height,
+		image:  C.make_image(C.int(width), C.int(height), 3),
+	}
+	C.fill_image_f32(&imgDarknet.image, C.int(width), C.int(height), 3, (*C.float)(unsafe.Pointer(&flatten[0])))
 	return imgDarknet, nil
 }

network.c

@@ -14,10 +14,8 @@ struct network_box_result perform_network_detect(network *n, image *img, int cla
         sized = resize_image(*img, n->w, n->h);
     }
     struct network_box_result result = { NULL };
-    float *X = sized.data;
+    network_predict(*n, sized.data);
-    network_predict(*n, X);
     int nboxes = 0;
-    detection *dets = get_network_boxes(n, img->w, img->h, thresh, hier_thresh, 0, 1, &nboxes, letter_box);
     result.detections = get_network_boxes(n, img->w, img->h, thresh, hier_thresh, 0, 1, &result.detections_len, letter_box);
     if (nms) {
         do_nms_sort(result.detections, result.detections_len, classes, nms);

network.go

@@ -71,8 +71,8 @@ func (n *YOLONetwork) Detect(img *DarknetImage) (*DetectionResult, error) {
 	startTime := time.Now()
 	result := C.perform_network_detect(n.cNet, &img.image, C.int(n.Classes), C.float(n.Threshold), C.float(n.hierarchalThreshold), C.float(n.nms), C.int(0))
 	endTime := time.Now()
-	defer C.free_detections(result.detections, result.detections_len)
 	ds := makeDetections(img, result.detections, int(result.detections_len), n.Threshold, n.Classes, n.ClassNames)
+	C.free_detections(result.detections, result.detections_len)
 	endTimeOverall := time.Now()
 	out := DetectionResult{
 		Detections:           ds,