test cuda9.0 on ubuntu 16.04

Author: springkim

CMakeLists.txt

@@ -143,15 +143,15 @@ if (OPENMP_FOUND)
     set (CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} ${OpenMP_EXE_LINKER_FLAGS}")
 endif()
 if(UNIX)
-	set(CUDNN /usr/local/cuda/lib64/libcudnn.so)
 	#set(CUDNN /usr/local/cuda/lib64/libcudnn.so)
+	#set(CUDNN usr/lib/aarch64-linux-gnu/libcudnn.so)
 	#set(CUDNN /usr/bin/libcudnn.so)
 	include_directories(/usr/local/cuda/include)
 	cuda_add_executable(YOLOv3SE_Train ${SOURCE_FILES} src/main.c)   #Make train executable
-	target_link_libraries(YOLOv3SE_Train cuda pthread m CUDNN)
+	target_link_libraries(YOLOv3SE_Train cuda pthread m ${CUDNN})
 	cuda_add_library(YOLOv3SE SHARED ${SOURCE_FILES} src/main.c src/yolo_v3.c)       #make so file
 	SET_TARGET_PROPERTIES(YOLOv3SE PROPERTIES COMPILE_FLAGS "-D_WINDLL")
-	target_link_libraries(YOLOv3SE cuda pthread m CUDNN)
+	target_link_libraries(YOLOv3SE cuda pthread m ${CUDNN})
 	target_include_directories(YOLOv3SE PUBLIC ${PROJECT_SOURCE_DIR}/include)
 endif(UNIX)
 if(MSVC)

README.md

@@ -1,26 +1,28 @@
-## YOLOv3_SpringEdition <img src="https://i.imgur.com/oYejfWp.png" title="Windows8" width="48">
+YOLOv3_SpringEdition <img src="https://i.imgur.com/oYejfWp.png" title="Windows8" width="48">
+--------------------------------------------------------------------------------------------
 
 <img src="https://i.imgur.com/ElCyyzT.png" title="Windows8" width="48"><img src="https://i.imgur.com/O5bye0l.png" width="48"><img src="https://i.imgur.com/kmfOMZz.png" width="48"><img src="https://i.imgur.com/6OT8yM9.png" width="48">
 
 #### YOLOv3 C++ Windows and Linux interface library. (Train,Detect both)
 
-* Remove pthread,opencv dependency.
-* You need only 1 files for YOLO deep-learning.
-* Support windows, linux as same interface.
+-	Remove pthread,opencv dependency.
+-	You need only 1 files for YOLO deep-learning.
+-	Support windows, linux as same interface.
 
 #### Do you want train YOLOv3 as double click? and detect using YOLOv3 as below?
+
 ```cpp
 YOLOv3 detector;
 detector.Create("coco.weights", "coco.cfg", "coco.names");
 cv::Mat img=cv::imread("a.jpg");
 std::vector<BoxSE> boxes = detector.Detect(img, 0.5F);
 ```
-* Then you've come to the right place.
+
+-	Then you've come to the right place.
 
 ### 1. Setup for train.
-You need only 2 files for train that are **YOLOv3SE_Train.exe** and **cudnn64_5.dll** on Windows.
-If you are on Linux, then you need only **YOLOv3SE_Train**.
-This files are in `YOLOv3_SpringEdition/bin`.
+
+You need only 2 files for train that are **YOLOv3SE_Train.exe** and **cudnn64_5.dll** on Windows. If you are on Linux, then you need only **YOLOv3SE_Train**. This files are in `YOLOv3_SpringEdition/bin`.
 
 The requirement interface not changed. Same as **[pjreddie/darknet](https://github.com/pjreddie/darknet)**.
 
@@ -34,58 +36,58 @@ And YOLOv3SE_Train.exe is automatically choosing multi-gpu training. and select
 
 ### 2. Setup for detect
 
-Just include **YOLOv3SE.h** and use it. See  `YOLOv3_SpringEdition_Test/`.
-You need only **YOLOv3SE.h**, **libYOLOv3SE.dll** and **cudnn64_5.dll** for detect.
+Just include **YOLOv3SE.h** and use it. See `YOLOv3_SpringEdition_Test/`. You need only **YOLOv3SE.h**, **libYOLOv3SE.dll** and **cudnn64_5.dll** for detect.
 
 ##### Reference
 
 The class `YOLOv3` that in `YOLOv3SE.h` has 3 methods.
+
 ```cpp
 void Create(std::string weights,std::string cfg,std::string names);
 ```
-This method load trained model(**weights**), network configuration(**cfg**) and class naming file(**names**)
-* **Parameter**
-	* **weights** : trained model path(e.g. "obj.weights")
-	* **cfg** : network configuration file(e.g. "obj.cfg")
-	* **names** : class naming file(e.g. "obj.names")
+
+This method load trained model(**weights**), network configuration(**cfg**) and class naming file(**names**\)* **Parameter** * **weights** : trained model path(e.g. "obj.weights") * **cfg** : network configuration file(e.g. "obj.cfg") * **names** : class naming file(e.g. "obj.names")
 
 ```cpp
 std::vector<BoxSE> Detect(cv::Mat img, float threshold);
 std::vector<BoxSE> Detect(std::string file, float threshold);
 std::vector<BoxSE> Detect(IplImage* img, float threshold);
 ```
-This method is detecting objects of `file`,`cv::Mat` or `IplImage`.
-* **Parameter**
-	* **file** : image file path
-	* **img** : 3-channel image.
-	* **threshold** : It removes predictive boxes if there score is less than threshold.
+
+This method is detecting objects of `file`,`cv::Mat` or `IplImage`.* **Parameter** * **file** : image file path * **img** : 3-channel image. * **threshold** : It removes predictive boxes if there score is less than threshold.
 
 ```cpp
 void Release();
 ```
+
 Release loaded network.
 
-## Technical issue
+Technical issue
+---------------
 
-Original YOLOv3(darknet) is linux version.
-And **[AlexeyAB](https://github.com/AlexeyAB/darknet)** already made YOLOv3 Windows version.
-But, his detection method is too slow on Windows. I don't know why exactly. Maybe it has bottleneck.
-So, I converted **[darknet](https://github.com/pjreddie/darknet)**(YOLOv3 only) again.
+Original YOLOv3(darknet) is linux version. And **[AlexeyAB](https://github.com/AlexeyAB/darknet)** already made YOLOv3 Windows version. But, his detection method is too slow on Windows. I don't know why exactly. Maybe it has bottleneck. So, I converted **[darknet](https://github.com/pjreddie/darknet)**(YOLOv3 only) again.
 
-## change log
+change log
+----------
 
 **build_windows.bat** and **build_linux.sh** will download automatically correct version of cudnn. and build as cmake.
+
 ```
-1080ti + CUDA8.0 + cudnn7.1 = 19FPS
-1080 + CUDA9.0 + cudnn7.1 = 27FPS
+Windows + 1080ti + CUDA8.0 + cudnn7.1 = 19FPS
+Windows + 1080   + CUDA9.0 + cudnn7.1 = 27FPS
+Ubuntu  + 1080   + CUDA8.0 + cudnn7.1 = 20FPS
+Ubuntu  + 1080   + CUDA9.0 + cudnn7.1 = 22FPS
 ```
-## Software requirement
 
-* CMake
-* CUDA 8.0 or 9.0 or 9.1
-* OpenCV
-* Visual Studio
+Software requirement
+--------------------
+
+-	CMake
+-	CUDA 8.0 or 9.0(9.1 is not working)
+-	OpenCV(for testing)
+-	Visual Studio
 
-## Hardware requirement
+Hardware requirement
+--------------------
 
-* NVIDIA GPU
+-	NVIDIA GPU

bin/CUDA8.0/YOLOv3SE_Train

@@ -1,4 +1,17 @@
 mkdir build
 cd build
-cmake ..
-make
+if [ -f "/usr/local/cuda/lib64/libcudnn.so" ]; then
+	#Default cuda path
+    cmake .. -DCUDNN=/usr/local/cuda/lib64/libcudnn.so
+	make
+elif [ -f "/usr/bin/libcudnn.so" ]; then
+	#global path
+    cmake .. -DCUDNN=/usr/bin/libcudnn.so
+	make
+elif [ -f "/usr/lib/aarch64-linux-gnu/libcudnn.so" ]; then
+	#jetson cuda path(jetpack)
+    cmake .. -DCUDNN=/usr/lib/aarch64-linux-gnu/libcudnn.so
+	make
+else
+	echo "there is no cudnn in your PC";
+fi