Add opencv script, modified macro, add opencv section into readme

Signed-off-by: Micaela Verucchi <micaela.verucchi@unimore.it>

README.md

@@ -4,6 +4,7 @@ The main goal of this project is to exploit NVIDIA boards as much as possible to
 
 ## Index
 * [Dependencies](#dependencies)
+* [About OpenCV](#about-opencv)
 * [How to compile this repo](#how-to-compile-this-repo)
 * [Workflow](#workflow)
 * [How to export weights](#how-to-export-weights)
@@ -23,6 +24,13 @@ This branch works on every NVIDIA GPU that supports the dependencies:
 * OPENCV 4.1
 * yaml-cpp 0.5.2 (sudo apt install libyaml-cpp-dev)
 
+## About OpenCV
+To compile and install OpenCV4 with contrib us the script ```install_OpenCV4.sh```. It will download and compile OpenCV in Download folder.
+```
+bash install_OpenCV4.sh
+```
+When using openCV not compiled with contrib, comment the definition of OPENCV_CUDACONTRIBCONTRIB in include/tkDNN/DetectionNN.h. When commented, the preprocessing of the networks is computed on the CPU, otherwise on the GPU. In the latter case some milliseconds are saved in the end-to-end latency. 
+
 ## How to compile this repo
 Build with cmake. If using Ubuntu 18.04 a new version of cmake is needed (1.15 or above). 
 ```

include/tkDNN/CenternetDetection.h

@@ -43,7 +43,7 @@ private:
     
     float *target_coords;
 
-    #ifdef OPENCV_CUDA
+    #ifdef OPENCV_CUDACONTRIB
         float *mean_d;
         float *stddev_d;
     #else

include/tkDNN/DetectionNN.h

@@ -17,7 +17,7 @@
 
 #include "tkdnn.h"
 
-#define OPENCV_CUDA //if OPENCV has been compiled with CUDA and contrib.
+#define OPENCV_CUDACONTRIB //if OPENCV has been compiled with CUDA and contrib.
 
 namespace tk { namespace dnn {
 
@@ -37,7 +37,7 @@ class DetectionNN {
 
         cv::Scalar colors[256];
 
-#ifdef OPENCV_CUDA
+#ifdef OPENCV_CUDACONTRIB
         cv::cuda::GpuMat bgr[3];
         cv::cuda::GpuMat imagePreproc;
 #else

src/CenternetDetection.cpp

@@ -91,7 +91,7 @@ bool CenternetDetection::init(const std::string& tensor_path, const int n_classe
 
     checkCuda( cudaMallocHost(&target_coords, 4 * K *sizeof(float)) );
 
-#ifdef OPENCV_CUDA
+#ifdef OPENCV_CUDACONTRIB
 
     checkCuda( cudaMalloc(&mean_d, 3 * sizeof(float)) );
     checkCuda( cudaMalloc(&stddev_d, 3 * sizeof(float)) );
@@ -177,7 +177,7 @@ void CenternetDetection::preprocess(cv::Mat &frame)
         // step_t = end_t;
     }
     sz_old = sz;
-#ifdef OPENCV_CUDA
+#ifdef OPENCV_CUDACONTRIB
     cv::cuda::GpuMat im_Orig; 
     cv::cuda::GpuMat imageF1_d, imageF2_d;
         

src/MobilenetDetection.cpp

@@ -160,7 +160,7 @@ bool MobilenetDetection::init(const std::string& tensor_path, const int n_classe
 
     generate_ssd_priors(specs, N_SSDSPEC);
 
-#ifndef OPENCV_CUDA
+#ifndef OPENCV_CUDACONTRIB
     checkCuda(cudaMallocHost(&input, sizeof(dnnType) * netRT->input_dim.tot()));
 #endif
     checkCuda(cudaMalloc(&input_d, sizeof(dnnType) * netRT->input_dim.tot()));
@@ -209,7 +209,7 @@ bool MobilenetDetection::init(const std::string& tensor_path, const int n_classe
 
 void MobilenetDetection::preprocess(cv::Mat &frame)
 {
-#ifdef OPENCV_CUDA
+#ifdef OPENCV_CUDACONTRIB
         //move original image on GPU
         cv::cuda::GpuMat orig_img, frame_nomean;
         orig_img = cv::cuda::GpuMat(frame);

src/Yolo3Detection.cpp

@@ -30,7 +30,7 @@ bool Yolo3Detection::init(const std::string& tensor_path, const int n_classes) {
     }
 
     dets = tk::dnn::Yolo::allocateDetections(tk::dnn::Yolo::MAX_DETECTIONS, classes);
-#ifndef OPENCV_CUDA
+#ifndef OPENCV_CUDACONTRIB
     checkCuda(cudaMallocHost(&input, sizeof(dnnType)*netRT->input_dim.tot()));
 #endif
     checkCuda(cudaMalloc(&input_d, sizeof(dnnType)*netRT->input_dim.tot()));
@@ -50,7 +50,7 @@ bool Yolo3Detection::init(const std::string& tensor_path, const int n_classes) {
 
 void Yolo3Detection::preprocess(cv::Mat &frame)
 {
-#ifdef OPENCV_CUDA
+#ifdef OPENCV_CUDACONTRIB
     cv::cuda::GpuMat orig_img, img_resized;
     orig_img = cv::cuda::GpuMat(frame);
     cv::cuda::resize(orig_img, img_resized, cv::Size(netRT->input_dim.w, netRT->input_dim.h));