mmr/tkDNN
Archived
1
0
Fork 0
Code Issues 46 Pull Requests 12 Releases 4 Activity

-Modified Dockerfile.base to cudagl

Browse Source 
-Changed demo to take in input from demoConfig.yaml file
-Readme changes for demo.md
This commit is contained in:
perseusdg
2021-11-24 03:39:57 +05:30
parent a8c98e3c31
commit 9cecc5051a
11 changed files with 422 additions and 124 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.gitignore
+1
View File
@@ -21,3 +21,4 @@ scripts/COCO_val2017/*
scripts/COCO_val2017.zip
scripts/all_labels.txt
/cmake/cuda_script
/cmake-build-debug/
demo/demo/demo.cpp
+55 -57
View File
@@ -18,64 +18,57 @@ void sig_handler(int signo) {
int main(int argc, char *argv[]) {
std::cout<<"detection\n";
signal(SIGINT, sig_handler);
#ifdef __linux__
std::string config_file = "../demo/demoConfig.yaml";
#elif _WIN32
std::string config_file = "..\\..\\..\\demo\\demoConfig.yaml";
#endif
std::string net = "yolo4tiny_fp32.rt";
#ifdef __linux__
std::string cfgPath = "../tests/darknet/cfg/yolo4tiny.cfg";
#elif _WIN32
std::string cfgPath = "..\\tests\\darknet\\cfg\\yolo4tiny.cfg";
#endif
#ifdef __linux__
std::string namePath = "../tests/darknet/names/coco.names";
#elif _WIN32
std::string namePath = "..\\tests\\darknet\\names\\coco.names";
#endif
if(argc > 1)
net = argv[1];
#ifdef __linux__
std::string input = "../demo/yolo_test.mp4";
#elif _WIN32
std::string input = "..\\demo\\yolo_test.mp4";
#endif
char ntype = 'y';
if(argc > 2)
input = argv[2];
if(argc > 3)
ntype = argv[3][0];
int n_classes = 80;
if(argc > 4)
n_classes = atoi(argv[4]);
if(argc > 5)
cfgPath = argv[5];
if(argc > 6)
namePath = argv[6];
int n_batch = 1;
if(argc > 7)
n_batch = atoi(argv[7]);
bool show = true;
if(argc > 8)
show = atoi(argv[8]);
float conf_thresh=0.3;
if(argc >= 9)
conf_thresh = atof(argv[9]);
if(n_batch < 1 || n_batch > 64)
FatalError("Batch dim not supported");
if(!show)
SAVE_RESULT = true;
if(ntype == 'c' || ntype == 'm'){
cfgPath = "";
namePath = "";
if(argc > 1){
config_file = argv[1];
}
YAML::Node conf = YAMLloadConf(config_file);
if(!conf){
FatalError("Problem with config file");
}
std::string net = YAMLgetConf<std::string>(conf,"net","yolo4tiny_fp32.rt");
if(!fileExist(net.c_str())) {
FatalError("The given network does not exist. Create the rt first.");
}
#ifdef __linux__
std::string input = YAMLgetConf<std::string>(conf, "input", "../demo/yolo_test.mp4");
std::string cfgPath = YAMLgetConf<std::string>(conf,"cfg_input", "../tests/darknet/cfg/yolo4tiny.cfg");
std::string namePath = YAMLgetConf<std::string>(conf,"name_input","../tests/darknet/names/coco.names");
#elif _WIN32
std::string input = YAMLgetConf(conf, "win_input", "..\\..\\..\\demo\\yolo_test.mp4");
std::string cfgPath = YAMLgetConf(conf,"cfg_win_input","..\\..\\..\\tests\\darknet\\cfg\\yolo4tiny.cfg");
std::string namePath = YAMLgetConf(conf,"name_win_input","..\\..\\..\\tests\\darknet\\names\\coco.names");
#endif
if(!fileExist(input.c_str()))
FatalError("The given input video does not exist.");
char ntype = YAMLgetConf<char>(conf, "ntype", 'y');
int n_classes = YAMLgetConf<int>(conf, "n_classes", 80);
int n_batch = YAMLgetConf<int>(conf, "n_batch", 1);
if(n_batch < 1 || n_batch > 64)
FatalError("Batch dim not supported");
float conf_thresh = YAMLgetConf<float>(conf, "conf_thresh", 0.3);
bool show = YAMLgetConf<bool>(conf, "show", true);
bool save = YAMLgetConf<bool>(conf, "save", false);
tk::dnn::Yolo3Detection yolo;
tk::dnn::CenternetDetection cnet;
tk::dnn::MobilenetDetection mbnet;
@@ -98,6 +91,11 @@ int main(int argc, char *argv[]) {
FatalError("Network type not allowed (3rd parameter)\n");
}
if(ntype == 'c' || ntype == 'm'){
cfgPath = "";
namePath = "";
}
detNN->init(net,cfgPath,namePath,n_classes,n_batch,conf_thresh);
gRun = true;
@@ -109,7 +107,7 @@ int main(int argc, char *argv[]) {
std::cout<<"camera started\n";
cv::VideoWriter resultVideo;
if(SAVE_RESULT) {
if(save) {
int w = cap.get(cv::CAP_PROP_FRAME_WIDTH);
int h = cap.get(cv::CAP_PROP_FRAME_HEIGHT);
resultVideo.open("result.mp4", cv::VideoWriter::fourcc('M','P','4','V'), 30, cv::Size(w, h));
@@ -149,7 +147,7 @@ int main(int argc, char *argv[]) {
cv::waitKey(1);
}
}
if(n_batch == 1 && SAVE_RESULT)
if(n_batch == 1 && save)
resultVideo << frame;
}
@@ -157,7 +155,7 @@ int main(int argc, char *argv[]) {
double mean = 0;
std::cout<<COL_GREENB<<"\n\nTime stats:\n";
std::cout<<"Min: "<<*std::min_element(detNN->stats.begin(), detNN->stats.end())/n_batch<<" ms\n";
std::cout<<"Min: "<<*std::min_element(detNN->stats.begin(), detNN->stats.end())/n_batch<<" ms\n";
std::cout<<"Max: "<<*std::max_element(detNN->stats.begin(), detNN->stats.end())/n_batch<<" ms\n";
for(int i=0; i<detNN->stats.size(); i++) mean += detNN->stats[i]; mean /= detNN->stats.size();
std::cout<<"Avg: "<<mean/n_batch<<" ms\t"<<1000/(mean/n_batch)<<" FPS\n"<<COL_END;
demo/demoConfig.yaml
+22
View File
@@ -0,0 +1,22 @@
# video input
input : "../demo/yolo_test.mp4"
win_input : "..\\..\\..\\demo\\yolo_test.mp4"
#cfg input
cfg_input : "../tests/darknet/cfg/yolo4tiny.cfg"
cfg_win_input : "..\\..\\..\\tests\\darknet\\cfg\\yolo4tiny.cfg"
#name input
name_input : "../tests/darknet/names/coco.names"
name_win_input : "..\\..\\..\\tests\\darknet\\names\\coco.names"
# network config
net : "yolo4tiny_fp32.rt"
ntype : 'y'
n_classes : 80
n_batch : 1
conf_thresh : 0.3
# demo config
show : true
save : false
docker/Dockerfile.base
+121 -38
View File
@@ -1,57 +1,140 @@
FROM nvidia/cuda:11.3.1-devel-ubuntu20.04
LABEL maintainer "Francesco Gatti"
ENV DEBIAN_FRONTEND=noninteractive
RUN apt-get update && apt-get install libcudnn8-dev=8.2.1.32-1+cuda11.3 libcudnn8=8.2.1.32-1+cuda11.3 libnvinfer-dev=8.0.3-1+cuda11.3 libnvinfer8=8.0.3-1+cuda11.3
RUN DEBIAN_FRONTEND=noninteractive apt-get update && apt install -y git wget libeigen3-dev libyaml-cpp-dev gcc-9 g++-9 libopengl-dev libgl-dev
RUN cd /tmp && \
wget https://github.com/Kitware/CMake/releases/download/v3.21.4/cmake-3.21.4-Linux-x86_64.sh && \
chmod +x cmake-3.21.4-Linux-x86_64.sh && \
./cmake-3.21.4-Linux-x86_64.sh --prefix=/usr/local --exclude-subdir --skip-license && \
rm ./cmake-3.21.4-Linux-x86_64.sh
FROM nvidia/cudagl:11.3.1-devel-ubuntu20.04
LABEL maintainer "TKDNN AUTHORS"
LABEL Description="tkDNN+cudagl"
LABEL com.tkdnn.nvidia.version="11.3.1"
ENV DEBIAN_FRONTEND noninteractive
ENV CC gcc
ENV CXX g++
RUN apt-get update && apt-get install -y \
libblkid-dev && apt-get clean && rm -rf /var/lib/apt/lists/*
RUN apt-get update && apt-get install -y \
libcudnn8-dev=8.2.1.32-1+cuda11.3 \
libcudnn8=8.2.1.32-1+cuda11.3 \
libnvinfer-dev=8.0.3-1+cuda11.3 \
libnvinfer8=8.0.3-1+cuda11.3 && apt-get clean && rm -rf /var/lib/apt/lists/*
RUN apt-get update && apt-get install -y --no-install-recommends \
libblkid-dev \
locales \
lsb-release \
mesa-utils \
git \
nano \
terminator \
wget \
curl \
libssl-dev \
htop \
dbus-x11 \
libqt5opengl5-dev \
libgtk-3-dev \
libvtk7-dev \
libv4l-dev \
tar \
libgoogle-glog-dev \
libgflags-dev \
gfortran-9 \
libtbb-dev \
libgstreamer1.0-dev \
libgstreamer-plugins-base1.0-dev \
libdc1394-22-dev \
libavresample-dev \
libatlas-cpp-0.6-dev \
python3-dev \
gdb \
python3-pip \
unzip libtbb-dev && \
apt-get clean && rm -rf /var/lib/apt/lists/*
RUN apt-get update && apt-get install -y --no-install-recommends \
software-properties-common && apt-get clean && rm -rf /var/lib/apt/lists/*
RUN apt-add-repository universe
RUN apt-get update && apt-get install -y python3-pip python3 openssh-server ssh pyqt5-dev sip-dev && apt-get clean && rm -rf /var/lib/apt/lists/*
RUN pip3 install --upgrade pip
RUN pip3 install --upgrade virtualenv
RUN pip3 install --upgrade paramiko
RUN pip3 install --ignore-installed --upgrade numpy protobuf
RUN cd ~ && mkdir build
RUN cd ~/build && wget https://github.com/Kitware/CMake/releases/download/v3.21.4/cmake-3.21.4.tar.gz && \
tar -xvf cmake-3.21.4.tar.gz && cd cmake-3.21.4 && ./configure --prefix=/usr/local --qt-gui --parallel=12 && \
make -j8 && make install
RUN apt-get update && apt-get install -y automake autoconf pkg-config libevent-dev libncurses5-dev bison && \
apt-get clean && rm -rf /var/lib/apt/lists/
RUN git clone https://github.com/tmux/tmux.git && \
cd tmux && git checkout tags/3.2 && ls -la && sh autogen.sh && ./configure && make -j8 && make install
RUN apt-get update && apt-get install -y zsh && apt-get clean && rm -rf /var/lib/apt/lists/*
RUN wget https://github.com/robbyrussell/oh-my-zsh/raw/master/tools/install.sh -O - | zsh || true
RUN chsh -s /usr/bin/zsh root
RUN git clone https://github.com/sindresorhus/pure /root/.oh-my-zsh/custom/pure
RUN ln -s /root/.oh-my-zsh/custom/pure/pure.zsh-theme /root/.oh-my-zsh/custom/
RUN ln -s /root/.oh-my-zsh/custom/pure/async.zsh /root/.oh-my-zsh/custom/
RUN sed -i -e 's/robbyrussell/refined/g' /root/.zshrc
RUN sed -i '/plugins=(/c\plugins=(git git-flow adb pyenv tmux)' /root/.zshrc
RUN mkdir -p /root/.config/terminator/
COPY assets/terminator_config /root/.config/terminator/config
RUN echo "/usr/local/nvidia/lib" >> /etc/ld.so.conf.d/nvidia.conf && \
echo "/usr/local/nvidia/lib64" >> /etc/ld.so.conf.d/nvidia.conf && \
echo "/usr/local/cuda/lib64" >> /etc/ld.so.conf.d/nvidia.conf
ENV PATH /usr/local/nvidia/bin:/usr/local/cuda/bin:${PATH}
ENV LD_LIBRARY_PATH /usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/lib:/usr/lib/x86_64-linux-gnu:/usr/local/lib:${LD_LIBRARY_PATH}
ENV NVIDIA_VISIBLE_DEVICES all
RUN echo "INSTALL OPENCV"
RUN apt-get install -y build-essential \
unzip \
pkg-config \
libjpeg-dev \
libpng-dev \
libtiff-dev \
libavcodec-dev \
libavformat-dev \
libswscale-dev \
libv4l-dev \
libxvidcore-dev \
libx264-dev \
libgtk-3-dev \
libatlas-base-dev \
gfortran-9 \
libtbb-dev \
libgstreamer1.0-dev \
libgstreamer-plugins-base1.0-dev \
libdc1394-22-dev \
libavresample-dev
RUN cd && wget https://github.com/opencv/opencv/archive/4.5.4.tar.gz && tar -xf 4.5.4.tar.gz && rm *.tar.gz
RUN cd && wget https://github.com/opencv/opencv_contrib/archive/4.5.4.tar.gz && tar -xf 4.5.4.tar.gz && rm *.tar.gz
RUN cd && \
ENV NVIDIA_DRIVER_CAPABILITIES compute,utility,graphics
RUN cd ~/build && wget https://github.com/opencv/opencv/archive/4.5.4.tar.gz && tar -xf 4.5.4.tar.gz && rm 4.5.4.tar.gz
RUN cd ~/build && wget https://github.com/opencv/opencv_contrib/archive/4.5.4.tar.gz && tar -xf 4.5.4.tar.gz && rm 4.5.4.tar.gz
RUN cd ~/build && \
cd opencv-4.5.4 && mkdir build && cd build && \
cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D INSTALL_PYTHON_EXAMPLES=OFF \
-D INSTALL_C_EXAMPLES=OFF \
-D OPENCV_EXTRA_MODULES_PATH='~/opencv_contrib-4.5.4/modules' \
-D OPENCV_EXTRA_MODULES_PATH='~/build/opencv_contrib-4.5.4/modules' \
-D BUILD_EXAMPLES=OFF \
-D BUILD_TESTS=OFF \
-D BUILD_PERF_TESTS=OFF \
-D BUILD_DOCS=OFF \
-D WITH_CUDA=ON \
-D WITH_OPENGL=ON \
-D WITH_NVCUVID=ON \
-D CUDA_ARCH_BIN=7.2 \
-D CUDA_ARCH_PTX="" \
-D CUDA_ARCH_PTX=7.2 \
-D ENABLE_FAST_MATH=ON \
-D CUDA_FAST_MATH=ON \
-D WITH_CUBLAS=ON \
-D WITH_CUDNN=ON \
-D WITH_OPENMP=ON \
-D WITH_NONFREE=ON \
-D WITH_LIBV4L=ON \
-D WITH_GSTREAMER=ON \
-D WITH_GSTREAMER_0_10=OFF \
-D WITH_TBB=ON \
../ && make -j12 && make install
RUN apt clean
../ && make -j12 && make install && ldconfig
RUN cd ~ && rm -rf build
RUN cd ~ && mkdir Development && cd Development && \
git clone https://github.com/ceccocats/tkDNN.git && cd tkDNN && \
mkdir build && cd build && \
cmake -DCMAKE_BUILD_TYPE=Release .. && \
make -j6
RUN apt-get clean && rm -rf /var/lib/apt/lists/*
COPY assets/entrypoint_setup.sh /
ENTRYPOINT ["/entrypoint_setup.sh"]
CMD ["terminator"]
docker/README.md
+1 -4
View File
@@ -9,13 +9,10 @@ docker build -t tkdnn:build -f Dockerfile .
# make nvidia docker working
# follow this guide: https://github.com/NVIDIA/nvidia-docker
# dowload tensorrt
# from: https://developer.nvidia.com/compute/machine-learning/tensorrt/secure/7.0/7.0.0.11/local_repo/nv-tensorrt-repo-ubuntu1804-cuda10.2-trt7.0.0.11-ga-20191216_1-1_amd64.deb
# build image
docker build -t ceccocats/tkdnn:latest -f Dockerfile.base .
# run image
docker run -ti --gpus all --rm ceccocats/tkdnn:latest bash
./docker_launch.sh
```
docker/assets/entrypoint_setup.sh
Executable
+123
View File
@@ -0,0 +1,123 @@
#! /bin/bash
CMD=
# Functions
# TOOD: Check if we can use: getent passwd $USER to extract all variables
# TODO: Check for valid inputs, cause now it will go through even with bad inputs
check_envs () {
DOCKER_CUSTOM_USER_OK=true;
if [ -z ${DOCKER_USER_NAME+x} ]; then
DOCKER_CUSTOM_USER_OK=false;
return;
fi
if [ -z ${DOCKER_USER_ID+x} ]; then
DOCKER_CUSTOM_USER_OK=false;
return;
else
if ! [ -z "${DOCKER_USER_ID##[0-9]*}" ]; then
echo -e "\033[1;33mWarning: User-ID should be a number. Falling back to defaults.\033[0m"
DOCKER_CUSTOM_USER_OK=false;
return;
fi
fi
if [ -z ${DOCKER_USER_GROUP_NAME+x} ]; then
DOCKER_CUSTOM_USER_OK=false;
return;
fi
if [ -z ${DOCKER_USER_GROUP_ID+x} ]; then
DOCKER_CUSTOM_USER_OK=false;
return;
else
if ! [ -z "${DOCKER_USER_GROUP_ID##[0-9]*}" ]; then
echo -e "\033[1;33mWarning: Group-ID should be a number. Falling back to defaults.\033[0m"
DOCKER_CUSTOM_USER_OK=false;
return;
fi
fi
}
setup_env_user () {
USER=$1
USER_ID=$2
GROUP=$3
GROUP_ID=$4
## Create user
useradd -m $USER
## Copy zsh/sh configs
cp /root/.profile /home/$USER/
cp /root/.bashrc /home/$USER/
cp /root/.zshrc /home/$USER/
## Copy terminator configs
mkdir -p /home/$USER/.config/terminator
cp /root/.config/terminator/config /home/$USER/.config/terminator/config
cp /root/.config/terminator/background.png /home/$USER/.config/terminator/background.png
cp -rf /root/.oh-my-zsh /home/$USER/
cp -rf /root/tkDNN /home/$USER/
rm -rf /home/$USER/.oh-my-zsh/custom/pure.zsh-theme /home/$USER/.oh-my-zsh/custom/async.zsh
ln -s /home/$USER/.oh-my-zsh/custom/pure/pure.zsh-theme /home/$USER/.oh-my-zsh/custom/
ln -s /home/$USER/.oh-my-zsh/custom/pure/async.zsh /home/$USER/.oh-my-zsh/custom/
sed -i -e 's@ZSH=\"/root@ZSH=\"/home/$USER@g' /home/$USER/.zshrc
# Copy SSH keys & fix owner
if [ -d "/root/.ssh" ]; then
cp -rf /root/.ssh /home/$USER/
chown -R $USER:$GROUP /home/$USER/.ssh
fi
## Fix owner
chown $USER:$GROUP /home/$USER
chown -R $USER:$GROUP /home/$USER/.config
chown $USER:$GROUP /home/$USER/.profile
chown $USER:$GROUP /home/$USER/.bashrc
chown $USER:$GROUP /home/$USER/.zshrc
chown -R $USER:$GROUP /home/$USER/.oh-my-zsh
chown -R $USER:$GROUP /home/$USER/tkDNN
## This a trick to keep the evnironmental variables of root which is important!
echo "if ! [ \"$DOCKER_USER_NAME\" = \"$(id -un)\" ]; then" >> /root/.bashrc
echo " cd /home/$DOCKER_USER_NAME" >> /root/.bashrc
echo " su $DOCKER_USER_NAME" >> /root/.bashrc
echo "fi" >> /root/.bashrc
echo "if ! [ \"$DOCKER_USER_NAME\" = \"$(id -un)\" ]; then" >> /root/.zshrc
echo " cd /home/$DOCKER_USER_NAME" >> /root/.zshrc
echo " su $DOCKER_USER_NAME" >> /root/.zshrc
echo "fi" >> /root/.zshrc
## Setup Password-file
PASSWDCONTENTS=$(grep -v "^${USER}:" /etc/passwd)
GROUPCONTENTS=$(grep -v -e "^${GROUP}:" -e "^docker:" /etc/group)
(echo "${PASSWDCONTENTS}" && echo "${USER}:x:$USER_ID:$GROUP_ID::/home/$USER:/bin/bash") > /etc/passwd
(echo "${GROUPCONTENTS}" && echo "${GROUP}:x:${GROUP_ID}:") > /etc/group
(if test -f /etc/sudoers ; then echo "${USER} ALL=(ALL) NOPASSWD: ALL" >> /etc/sudoers ; fi)
}
# ---Main---
# Create new user
## Check Inputs
check_envs
## Determine user & Setup Environment
if [ $DOCKER_CUSTOM_USER_OK == true ]; then
echo " -->DOCKER_USER Input is set to '$DOCKER_USER_NAME:$DOCKER_USER_ID:$DOCKER_USER_GROUP_NAME:$DOCKER_USER_GROUP_ID'";
echo -e "\033[0;32mSetting up environment for user=$DOCKER_USER_NAME\033[0m"
setup_env_user $DOCKER_USER_NAME $DOCKER_USER_ID $DOCKER_USER_GROUP_NAME $DOCKER_USER_GROUP_ID
else
echo " -->DOCKER_USER* variables not set. Using 'root'.";
echo -e "\033[0;32mSetting up environment for user=root\033[0m"
DOCKER_USER_NAME="root"
fi
# Change shell to zsh
chsh -s /usr/bin/zsh $DOCKER_USER_NAME
# Run CMD from Docker
"$@"
docker/assets/terminator_config
+18
View File
@@ -0,0 +1,18 @@
[global_config]
title_transmit_bg_color = "#2e3436"
[keybindings]
[layouts]
[[default]]
[[[child1]]]
parent = window0
type = Terminal
[[[window0]]]
parent = ""
type = Window
[plugins]
[profiles]
[[default]]
background_color = "#282828"
cursor_color = "#aaaaaa"
foreground_color = "#f3f3f3"
palette = "#000000:#aa0000:#00aa00:#c4a000:#3465a4:#75507b:#06989a:#d3d7cf:#88807c:#f15d22:#73c48f:#ffce51:#48b9c7:#ad7fa8:#34e2e2:#eeeeec"
docker/docker_launch.sh
Executable
+9
View File
@@ -0,0 +1,9 @@
xhost local:root
docker run --rm -it --runtime=nvidia --privileged --net=host --cap-add sys_ptrace -d --ipc=host \
-v /tmp/.X11-unix:/tmp/.X11-unix -e DISPLAY=$DISPLAY \
-v $HOME/.Xauthority:/home/$(id -un)/.Xauthority -e XAUTHORITY=/home/$(id -un)/.Xauthority \
-e DOCKER_USER_NAME=$(id -un) \
-e DOCKER_USER_ID=$(id -u) \
-e DOCKER_USER_GROUP_NAME=$(id -gn) \
-e DOCKER_USER_GROUP_ID=$(id -g) \
-v $HOME/.ssh:/home/$(id -un)/.ssh ceccocats/tkdnn
docs/demo.md
+19 -24
View File
@@ -30,31 +30,24 @@ cmake .. -DCMAKE_BUILD_TYPE=Debug -DDEBUG=True
make
```
Once you have successfully created your rt file, run the demo(yolo) :
Once you have successfully created your rt file, run the demo:
```
./demo yolo4_fp32.rt ../demo/yolo_test.mp4 y 80 ../tests/darknet/cfg/yolo4.cfg ../tests/darknet/names/coco.names
./ demo <path-to-config>
```
In general the demo program takes 1 parameter, the ```<path-to-config>``` that is the path to che configuration file. The parameter is optional and its default value is ```"../demo/demoConfig.yaml"```.
To run demo for mobilenet and centernet for the created rt file :
```
./demo mobilenetv2ssd_fp32.rt m 20
```
In general the demo program takes 7 parameters:
```
./demo <network-rt-file> <path-to-video> <kind-of-network> <number-of-classes> <cfg-path> <name-path> <n-batches> <show-flag> <conf-thresh>
```
where
* ```<network-rt-file>``` is the rt file generated by a test
* ```<<path-to-video>``` is the path to a video file or a camera input
* ```<kind-of-network>``` is the type of network. Thee types are currently supported: ```y``` (YOLO family), ```c``` (CenterNet family) and ```m``` (MobileNet-SSD family)
* ```<number-of-classes>```is the number of classes the network is trained on
* ```<cfg-path> ```is the relative path to the config file (only for darknet based networks) used to train the network
* ```<name-path>```is the relative path to the names file (only for darknet based networks) used to train the network
* ```<n-batches>``` number of batches to use in inference (N.B. you should first export TKDNN_BATCHSIZE to the required n_batches and create again the rt file for the network).
* ```<show-flag>``` if set to 0 the demo will not show the visualization but save the video into result.mp4 (if n-batches ==1)
* ```<conf-thresh>``` confidence threshold for the detector. Only bounding boxes with threshold greater than conf-thresh will be displayed.
The config file is a yaml file with the following attributes:
* ```net``` is the rt file generated by a test
* ```input``` is the path to a video file or a camera input (on Linux)
* ```win_input``` is the path to a video file or a camera input (on Windows)
* ```ntype``` is the type of network. Thee types are currently supported: ```y``` (YOLO family), ```c``` (CenterNet family) and ```m``` (MobileNet-SSD family)
* ```n_classes``` is the number of classes the network is trained on
* ```n_batch``` number of batches to use in inference (N.B. you should first export TKDNN_BATCHSIZE to the required n_batches and create again the rt file for the network).
* ```conf_thresh``` confidence threshold for the detector. Only bounding boxes with threshold greater than conf-thresh will be displayed.
* ```show``` if set to 0 the demo will not show the visualization (if n-batches ==1)
* ```save``` if set to 1 the demo will save the video of the demo into result.mp4 (if n-batches ==1)
* ```cfg_input``` (for linux) \ ```cfg_win_input``` (for windows) is the location of the cfg path of the network for mobilenet and centernet networks use ```" "```
* ```name_input``` (for linux) \ ```name_win_input``` (for windows) is the location of the name path of the network for mobilenet and centernet networks use ```" "```
N.B. By default it is used FP32 inference
@@ -69,7 +62,8 @@ To run the demo with FP16 inference follow these steps (example with yolov3):
export TKDNN_MODE=FP16 # set the half floating point optimization
rm yolo4_fp16.rt # be sure to delete(or move) old tensorRT files
./test_yolo4 # run the yolo test (is slow)
./demo yolo4_fp16.rt ../demo/yolo_test.mp4 y 80 ../tests/darknet/cfg/yolo4.cfg ../tests/darknet/names/coco.names
#set net: yolo4_fp16.rt in the config file
./demo
```
N.B. Using FP16 inference will lead to some errors in the results (first or second decimal).
@@ -94,7 +88,8 @@ export TKDNN_CALIB_LABEL_PATH=../demo/COCO_val2017/all_labels.txt
export TKDNN_CALIB_IMG_PATH=../demo/COCO_val2017/all_images.txt
rm yolo4_int8.rt # be sure to delete(or move) old tensorRT files
./test_yolo4 # run the yolo test (is slow)
./demo yolo4_int8.rt ../demo/yolo_test.mp4 y 80 ../tests/darknet/cfg/yolo4.cfg ../tests/darknet/names/coco.names
#set net: yolo4_int8.rt in the config file
./demo
```
N.B.
include/tkDNN/utils.h
+16 -1
View File
@@ -6,6 +6,8 @@
#include <fstream>
#include <iomanip>
#include <stdlib.h>
#include <yaml-cpp/yaml.h>
#include "cuda.h"
#include "cuda_runtime_api.h"
@@ -16,7 +18,6 @@
#ifdef __linux__
#include <unistd.h>
#endif
#include <ios>
@@ -161,5 +162,19 @@ static inline bool isCudaPointer(void *data) {
return cudaPointerGetAttributes(&attr, data) == 0;
}
inline YAML::Node YAMLloadConf(const std::string& conf_file) {
std::cerr<<"Loading YAML: "<<conf_file<<"\n";
return YAML::LoadFile(conf_file);
}
template<typename T>
inline T YAMLgetConf(YAML::Node conf, std::string key, T defaultVal) {
T val = defaultVal;
if(conf && conf[key]) {
val = conf[key].as<T>();
}
return val;
}
#endif //UTILS_H
scripts/checkExecTimes.py
+37
View File
@@ -0,0 +1,37 @@
import sys
import pandas as pd
if len(sys.argv) < 3:
print("Error: two csv files are needed, old first new second")
exit(1)
old_perf_file = str(sys.argv[1])
new_perf_file = str(sys.argv[2])
verbose = False
if len(sys.argv) == 4:
verbose = bool(sys.argv[3])
print("Comparing {} vs {}".format(old_perf_file, new_perf_file))
df_old = pd.read_csv (old_perf_file, sep=';', header=None, index_col=0)
df_new = pd.read_csv (new_perf_file, sep=';', header=None, index_col=0)
for index, row in df_new.iterrows():
if index in df_old.index:
if verbose:
print("New: ",row[1], row[2], row[3])
print("Old: ",df_old.loc[index][1], df_old.loc[index][2], df_old.loc[index][3])
print(index, end=': ')
if abs(row[1] - df_old.loc[index][1]) < df_old.loc[index][1]*0.1:
print("similar performance")
elif (row[1] < df_old.loc[index][1]):
print('\x1b[3;30;42m' + 'faster' + '\x1b[0m')
elif (row[1] > df_old.loc[index][1]):
if row[1] > df_old.loc[index][1] + df_old.loc[index][1] * 0.5 :
print('\x1b[3;30;41m' + 'WAY SLOWER' + '\x1b[0m')
else:
print('\x1b[3;30;41m' + 'slower' + '\x1b[0m')