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

darknet parser to be tested on yolo3

Browse Source
This commit is contained in:
Francesco Gatti
2020-05-30 19:38:26 +02:00
parent 298487d5c5 e5e6654b1d
commit 0ae96a6bc4
2 changed files with 123 additions and 77 deletions
Download Patch File Download Diff File Expand all files Collapse all files
include/tkDNN/DarknetParser.h
+105 -39
View File
@@ -8,27 +8,28 @@ namespace tk { namespace dnn {
std::string type = "";
int width = 0;
int height = 0;
int channels = 0;
int channels = 3;
int batch_normalize=0;
int groups = 0;
int filters=0;
int size_x=0;
int size_y=0;
int stride_x=0;
int stride_y=0;
int groups = 1;
int filters=1;
int size_x=1;
int size_y=1;
int stride_x=1;
int stride_y=1;
int padding_x = 0;
int padding_y = 0;
int n_mask = 0;
int classes = 0;
int num = 0;
float scale_xy = 0;
int classes = 20;
int num = 1;
int pad = 0;
float scale_xy = 1;
std::vector<int> layers;
std::string activation = "";
std::string activation = "linear";
};
std::ostream& operator<<(std::ostream& os, const darknetFields_t& f){
os << f.width << " " << f.height << " " << f.channels << " " << f.batch_normalize<< " " << f.filters << " " << " " << f.activation;
os << f.width << " " << f.height << " " << f.channels << " " << f.batch_normalize<< " " << f.filters << " " << f.activation<< " " << f.scale_xy;
return os;
}
@@ -52,26 +53,68 @@ namespace tk { namespace dnn {
return true;
}
std::vector<int> fromStringToIntVec(const std::string& line, const char delimiter){
std::stringstream linestream(line);
std::string value;
std::vector<int> values;
while(getline(linestream,value,delimiter))
values.push_back(std::stoi(value));
return values;
}
bool darknetParseFields(const std::string& line, darknetFields_t& fields){
std::string name,value;
if(!divideNameAndValue(line, name, value))
return false;
//std::cout<<name<<std::endl;
//std::cout<<value<<std::endl;
if(name == "width")
if(name.find("width") != std::string::npos)
fields.width = std::stoi(value);
else if (name == "height")
else if(name.find("height") != std::string::npos)
fields.height = std::stoi(value);
else if (name == "channels")
else if(name.find("channels") != std::string::npos)
fields.channels = std::stoi(value);
else if (name == "batch_normalize")
else if(name.find("batch_normalize") != std::string::npos)
fields.batch_normalize = std::stoi(value);
else if (name == "filters")
else if(name.find("filters") != std::string::npos)
fields.filters = std::stoi(value);
else if (name == "activation")
else if(name.find("activation") != std::string::npos)
fields.activation = value;
else if(name.find("size") != std::string::npos){
fields.size_x = std::stoi(value);
fields.size_y = std::stoi(value);
}
else if(name.find("size_x") != std::string::npos)
fields.size_x = std::stoi(value);
else if(name.find("size_y") != std::string::npos)
fields.size_y = std::stoi(value);
else if(name.find("stride") != std::string::npos){
fields.stride_x = std::stoi(value);
fields.stride_y = std::stoi(value);
}
else if(name.find("stride_x") != std::string::npos)
fields.stride_x = std::stoi(value);
else if(name.find("stride_y") != std::string::npos)
fields.stride_y = std::stoi(value);
else if(name.find("pad") != std::string::npos)
fields.pad = std::stoi(value);
else if(name.find("classes") != std::string::npos)
fields.classes = std::stoi(value);
else if(name.find("num") != std::string::npos)
fields.num = std::stoi(value);
else if(name.find("scale_xy") != std::string::npos)
fields.scale_xy = std::stof(value);
else if(name.find("from") != std::string::npos)
fields.layers.push_back(std::stof(value));
else if(name.find("mask") != std::string::npos){
auto vec = fromStringToIntVec(value, ',');
fields.n_mask = vec.size();
}
else if(name.find("layers") != std::string::npos)
fields.layers = fromStringToIntVec(value, ',');
else
std::cout<<"Not supported field: "<<line<<std::endl;
return true;
}
@@ -81,40 +124,59 @@ namespace tk { namespace dnn {
return new tk::dnn::Network(dim);
}
void darknetAddLayer(tk::dnn::Network *net, darknetFields_t &f, std::string wgs_path) {
void darknetAddLayer(tk::dnn::Network *net, darknetFields_t &f, std::string wgs_path, std::vector<tk::dnn::Layer*> &netLayers) {
if(net == nullptr)
FatalError("Cant add a layer without a Net\n");
// padding compute
if(f.pad == 1) {
f.padding_x = f.padding_y = f.size_x /2;
}
std::cout<<"Add layer: "<<f.type<<"\n";
if(f.type == "convolutional") {
std::string wgs = wgs_path + "/c" + std::to_string(net->num_layers) + ".bin";
std::string wgs = wgs_path + "/c" + std::to_string(netLayers.size()) + ".bin";
printf("%d (%d,%d) (%d,%d) (%d,%d) %s %d %d\n", f.filters, f.size_x, f.size_y, f.stride_x, f.stride_y, f.padding_x, f.padding_y, wgs.c_str(), f.batch_normalize, f.groups);
new tk::dnn::Conv2d(net, f.filters, f.size_x, f.size_y, f.stride_x,
f.stride_y, f.padding_x, f.padding_y, wgs, f.batch_normalize, false, f.groups);
netLayers.push_back(new tk::dnn::Conv2d(net, f.filters, f.size_x, f.size_y, f.stride_x,
f.stride_y, f.padding_x, f.padding_y, wgs, f.batch_normalize, false, f.groups));
if(f.activation != "linear") {
tkdnnActivationMode_t act;
if(f.activation == "relu") act = tkdnnActivationMode_t(CUDNN_ACTIVATION_RELU);
else if(f.activation == "leaky") act = tk::dnn::ACTIVATION_LEAKY;
else if(f.activation == "mish") act = tk::dnn::ACTIVATION_MISH;
else { FatalError("activation not supported: " + f.activation); }
new tk::dnn::Activation(net, act);
}
} else if(f.type == "shortcut") {
if(f.layers.size() != 1) FatalError("no layers to shortcut\n");
int layerIdx = net->num_layers + f.layers[0];
if(layerIdx < 0 || layerIdx >= net->num_layers) FatalError("impossible to shortcut\n");
std::cout<<"shortcut to "<<layerIdx<<" "<<net->layers[layerIdx]->getLayerName()<<"\n";
new tk::dnn::Shortcut(net, net->layers[layerIdx]);
int layerIdx = f.layers[0];
if(layerIdx < 0)
layerIdx = netLayers.size() + layerIdx;
if(layerIdx < 0 || layerIdx >= netLayers.size()) FatalError("impossible to shortcut\n");
std::cout<<"shortcut to "<<layerIdx<<" "<<netLayers[layerIdx]->getLayerName()<<"\n";
netLayers.push_back(new tk::dnn::Shortcut(net, netLayers[layerIdx]));
} else if(f.type == "upsample") {
new tk::dnn::Upsample(net, f.stride_x);
netLayers.push_back(new tk::dnn::Upsample(net, f.stride_x));
} else if(f.type == "route") {
if(f.layers.size() == 0) FatalError("no layers to Route\n");
std::vector<tk::dnn::Layer*> layers;
for(int i=0; i<f.layers.size(); i++) {
int layerIdx = net->num_layers + f.layers[i];
if(layerIdx < 0 || layerIdx >= net->num_layers) FatalError("impossible to shortcut\n");
layers.push_back(net->layers[layerIdx]);
int layerIdx = f.layers[i];
if(layerIdx < 0)
layerIdx = netLayers.size() + layerIdx;
if(layerIdx < 0 || layerIdx >= netLayers.size()) FatalError("impossible to route\n");
std::cout<<"Route to "<<layerIdx<<" "<<netLayers[layerIdx]->getLayerName()<<"\n";
layers.push_back(netLayers[layerIdx]);
}
new tk::dnn::Route(net, layers.data(), layers.size());
netLayers.push_back(new tk::dnn::Route(net, layers.data(), layers.size()));
} else if(f.type == "yolo") {
std::string wgs = wgs_path + "/g" + std::to_string(net->num_layers) + ".bin";
new tk::dnn::Yolo(net, f.classes, f.num, wgs, f.n_mask, f.scale_xy);
std::string wgs = wgs_path + "/g" + std::to_string(netLayers.size()) + ".bin";
printf("%d %d %s %d %f\n", f.classes, f.num/f.n_mask, wgs.c_str(), f.n_mask, f.scale_xy);
netLayers.push_back(new tk::dnn::Yolo(net, f.classes, f.num/f.n_mask, wgs, f.n_mask, f.scale_xy));
} else{
FatalError("layer not supported: " + f.type);
@@ -124,6 +186,9 @@ namespace tk { namespace dnn {
tk::dnn::Network* darknetParser(std::string cfg_file, std::string wgs_path) {
tk::dnn::Network *net = nullptr;
// layers without activations to retrive correct id number
std::vector<tk::dnn::Layer*> netLayers;
std::ifstream if_cfg(cfg_file);
if(!if_cfg.is_open())
@@ -149,7 +214,7 @@ namespace tk { namespace dnn {
if(fields.type == "net")
net = darknetAddNet(fields);
else
darknetAddLayer(net, fields, wgs_path);
darknetAddLayer(net, fields, wgs_path, netLayers);
}
// new type
@@ -168,12 +233,13 @@ namespace tk { namespace dnn {
// end of filled type
if(fields.type != "") {
darknetAddLayer(net, fields, wgs_path);
darknetAddLayer(net, fields, wgs_path, netLayers);
}
if(net == nullptr) {
FatalError("net not found\n");
}
return net;
}
tests/yolo3/yolo3.cpp
+18 -38
View File
@@ -11,72 +11,52 @@ int main() {
tk::dnn::Network *net = tk::dnn::darknetParser("../tests/yolo3/yolov3.cfg", "yolo3/layers");
net->print();
/*
int classes = 80;
tk::dnn::Yolo *yolo [3];
#include "models/Yolo3.h"
std::vector<tk::dnn::Yolo*> yolo;
for(int i=0; i<net->num_layers; i++) {
if(net->layers[i]->getLayerType() == tk::dnn::layerType_t::LAYER_YOLO)
yolo.push_back((tk::dnn::Yolo*)net->layers[i]);
}
// fill classes names
for(int i=0; i<3; i++) {
yolo[i]->classesNames = {"person" , "bicycle" , "car" , "motorbike" , "aeroplane" , "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" , "sofa" , "pottedplant" , "bed" , "diningtable" , "toilet" , "tvmonitor" , "laptop" , "mouse" , "remote" , "keyboard" , "cell phone" , "microwave" , "oven" , "toaster" , "sink" , "refrigerator" , "book" , "clock" , "vase" , "scissors" , "teddy bear" , "hair drier" , "toothbrush"};
}
std::string input_bin = bin_path + "/layers/input.bin";
std::vector<std::string> output_bins = {
bin_path + "/debug/layer82_out.bin",
bin_path + "/debug/layer94_out.bin",
bin_path + "/debug/layer106_out.bin"
};
// Load input
dnnType *data;
dnnType *input_h;
readBinaryFile(input_bin, dim.tot(), &input_h, &data);
//print network model
net.print();
readBinaryFile(input_bin, net->input_dim.tot(), &input_h, &data);
//convert network to tensorRT
tk::dnn::NetworkRT netRT(&net, net.getNetworkRTName("yolo3"));
tk::dnn::NetworkRT netRT(net, net->getNetworkRTName("yolo3"));
// the network have 3 outputs
tk::dnn::dataDim_t out_dim[3];
for(int i=0; i<3; i++) out_dim[i] = yolo[i]->output_dim;
dnnType *cudnn_out[3], *rt_out[3];
tk::dnn::dataDim_t dim1 = dim; //input dim
tk::dnn::dataDim_t dim1 = net->input_dim; //input dim
printCenteredTitle(" CUDNN inference ", '=', 30); {
dim1.print();
TIMER_START
net.infer(dim1, data);
net->infer(dim1, data);
TIMER_STOP
dim1.print();
}
for(int i=0; i<3; i++) cudnn_out[i] = yolo[i]->dstData;
printCenteredTitle(" compute detections ", '=', 30);
TIMER_START
int ndets = 0;
tk::dnn::Yolo::detection *dets = tk::dnn::Yolo::allocateDetections(tk::dnn::Yolo::MAX_DETECTIONS, classes);
for(int i=0; i<3; i++) yolo[i]->computeDetections(dets, ndets, net.input_dim.w, net.input_dim.h, 0.5);
tk::dnn::Yolo::mergeDetections(dets, ndets, classes);
for(int j=0; j<ndets; j++) {
tk::dnn::Yolo::box b = dets[j].bbox;
int x0 = (b.x-b.w/2.);
int x1 = (b.x+b.w/2.);
int y0 = (b.y-b.h/2.);
int y1 = (b.y+b.h/2.);
int cl = 0;
for(int c = 0; c < classes; ++c){
float prob = dets[j].prob[c];
if(prob > 0)
cl = c;
}
std::cout<<cl<<": "<<x0<<" "<<y0<<" "<<x1<<" "<<y1<<"\n";
}
TIMER_STOP
tk::dnn::dataDim_t dim2 = dim;
tk::dnn::dataDim_t dim2 = net->input_dim;
printCenteredTitle(" TENSORRT inference ", '=', 30); {
dim2.print();
TIMER_START
@@ -100,5 +80,5 @@ int main() {
ret_cudnn_tensorrt |= checkResult(odim, cudnn_out[i], rt_out[i]) == 0 ? 0 : ERROR_CUDNNvsTENSORRT;
}
return ret_cudnn | ret_tensorrt | ret_cudnn_tensorrt;
*/
}