HI all,
I am trying to run the converted caffe model in c++ code. But cannot get the same results as in demo_caffe.py.
Anyone tried this? Thanks.
Following is my code, modified based on ssd_detect.cpp
// This is a demo code for using a SSD model to do detection.
// The code is modified from examples/cpp_classification/classification.cpp.
// Usage:
// ssd_detect [FLAGS] model_file weights_file list_file
//
// where model_file is the .prototxt file defining the network architecture, and
// weights_file is the .caffemodel file containing the network parameters, and
// list_file contains a list of image files with the format as follows:
// folder/img1.JPEG
// folder/img2.JPEG
// list_file can also contain a list of video files with the format as follows:
// folder/video1.mp4
// folder/video2.mp4
//
#include <caffe/caffe.hpp>
#ifdef USE_OPENCV
#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgproc/imgproc.hpp>
// using namespace cv;
#endif // USE_OPENCV
#include
#include
#include
#include
#include
#include
#include
#include
// using namespace std;
using namespace caffe; // NOLINT(build/namespaces)
class Detector {
public:
Detector(const string& model_file,
const string& weights_file);
std::vector<vector > Detect(const cv::Mat& img);
private:
void SetMean(const string& mean_file, const string& mean_value);
void WrapInputLayer(std::vectorcv::Mat* input_channels);
void Preprocess(const cv::Mat& img,
std::vectorcv::Mat* input_channels);
private:
shared_ptr<Net > net_;
cv::Size input_geometry_;
int num_channels_;
// cv::Mat mean_;
};
Detector::Detector(const string& model_file,
const string& weights_file) {
#ifdef CPU_ONLY
Caffe::set_mode(Caffe::CPU);
#else
Caffe::set_mode(Caffe::GPU);
#endif
/* Load the network. */
net_.reset(new Net(model_file, TEST));
net_->CopyTrainedLayersFrom(weights_file);
CHECK_EQ(net_->num_inputs(), 1) << "Network should have exactly one input.";
CHECK_EQ(net_->num_outputs(), 1) << "Network should have exactly one output.";
Blob* input_layer = net_->input_blobs()[0];
num_channels_ = input_layer->channels();
CHECK(num_channels_ == 3 || num_channels_ == 1)
<< "Input layer should have 1 or 3 channels.";
input_geometry_ = cv::Size(input_layer->width(), input_layer->height());
}
std::vector<vector > Detector::Detect(const cv::Mat& img) {
Blob* input_layer = net_->input_blobs()[0];
input_layer->Reshape(1, num_channels_,
input_geometry_.height, input_geometry_.width);
/* Forward dimension change to all layers. */
net_->Reshape();
std::vectorcv::Mat input_channels;
WrapInputLayer(&input_channels);
Preprocess(img, &input_channels);
net_->Forward();
/* Copy the output layer to a std::vector /
Blob result_blob = net_->output_blobs()[0];
const float* result = result_blob->cpu_data();
const int num_det = result_blob->height();
vector<vector > detections;
for (int k = 0; k < num_det; ++k) {
if (result[0] == -1) {
// Skip invalid detection.
result += 7;
continue;
}
vector detection(result, result + 7);
detections.push_back(detection);
result += 7;
}
return detections;
}
/* Wrap the input layer of the network in separate cv::Mat objects
- (one per channel). This way we save one memcpy operation and we
- don't need to rely on cudaMemcpy2D. The last preprocessing
- operation will write the separate channels directly to the input
- layer. /
void Detector::WrapInputLayer(std::vectorcv::Mat input_channels) {
Blob* input_layer = net_->input_blobs()[0];
int width = input_layer->width();
int height = input_layer->height();
float* input_data = input_layer->mutable_cpu_data();
for (int i = 0; i < input_layer->channels(); ++i) {
cv::Mat channel(height, width, CV_32FC1, input_data);
input_channels->push_back(channel);
input_data += width * height;
}
}
void Detector::Preprocess(const cv::Mat& img,
std::vectorcv::Mat* input_channels) {
/* Convert the input image to the input image format of the network. */
cv::Mat sample;
if (img.channels() == 3 && num_channels_ == 1)
cv::cvtColor(img, sample, cv::COLOR_BGR2GRAY);
else if (img.channels() == 4 && num_channels_ == 1)
cv::cvtColor(img, sample, cv::COLOR_BGRA2GRAY);
else if (img.channels() == 4 && num_channels_ == 3)
cv::cvtColor(img, sample, cv::COLOR_BGRA2BGR);
else if (img.channels() == 1 && num_channels_ == 3)
cv::cvtColor(img, sample, cv::COLOR_GRAY2BGR);
else
sample = img;
cv::Mat sample_resized;
if (sample.size() != input_geometry_)
cv::resize(sample, sample_resized, input_geometry_);
else
sample_resized = sample;
// cv::cvtColor(sample_resized, sample_resized, CV_BGR2RGB);
cv::Mat sample_float;
if (num_channels_ == 3)
sample_resized.convertTo(sample_float, CV_32FC3);
else
sample_resized.convertTo(sample_float, CV_32FC1);
// cv::Mat sample_normalized;
// cv::subtract(sample_float, mean_, sample_normalized);
//
// normalize image
float *im_data = (float*)sample_float.data;
for(int i = 0; i < num_channels_ * sample_float.rows * sample_float.cols; i ++){
im_data[i] -= 127.5;
im_data[i] /= 127.5;
}
/* This operation will write the separate BGR planes directly to the
- input layer of the network because it is wrapped by the cv::Mat
- objects in input_channels. */
cv::split(sample_float, *input_channels);
CHECK(reinterpret_cast<float*>(input_channels->at(0).data)
== net_->input_blobs()[0]->cpu_data())
<< "Input channels are not wrapping the input layer of the network.";
}
DEFINE_string(mean_file, "",
"The mean file used to subtract from the input image.");
DEFINE_string(mean_value, "127.5", // 104,117,123
"If specified, can be one value or can be same as image channels"
" - would subtract from the corresponding channel). Separated by ','."
"Either mean_file or mean_value should be provided, not both.");
DEFINE_string(file_type, "image",
"The file type in the list_file. Currently support image and video.");
DEFINE_string(out_file, "",
"If provided, store the detection results in the out_file.");
DEFINE_double(confidence_threshold, 0.01,
"Only store detections with score higher than the threshold.");
static std::string SplitFilename (const std::string& str)
{
size_t found;
// std::cout << "Splitting: " << str << std::endl;
found=str.find_last_of ( "/\" );
// string rawname = fullname.substr(0, lastindex);
// std::cout << " folder: " << str.substr(0,found) << std::endl;
// std::cout << " file: " << str.substr(found+1) << std::endl;
std::string fname = str.substr ( found+1);
size_t lastindex = fname.find_last_of ( "." );
return fname.substr(0, lastindex);
}
static std::string gen_name(std::string dir, string fname){
return dir + "/" + fname + "_mnet2_cf.jpg";
}
int main(int argc, char** argv)
{
const string& model_file = argv[1];
const string& weights_file = argv[2];
const string& in_file = argv[3];
string fname = SplitFilename(in_file);
const string& out_dir = argv[4];
string out_file = gen_name(out_dir, fname);
const float confidence_threshold = FLAGS_confidence_threshold;
// Initialize the network.
Detector detector ( model_file, weights_file );
cv::Mat img = cv::imread ( in_file, -1 );
CHECK ( !img.empty() ) << "Unable to decode image " << in_file;
std::vector<vector<float> > detections = detector.Detect ( img );
/* Print the detection results. */
for ( int i = 0; i < detections.size(); ++i ) {
const vector<float>& d = detections[i];
// Detection format: [image_id, label, score, xmin, ymin, xmax, ymax].
CHECK_EQ ( d.size(), 7 );
const float score = d[2];
if ( score >= confidence_threshold ) {
std::cout << in_file << " ";
std::cout << static_cast<int> ( d[1] ) << " ";
std::cout << score << " ";
std::cout << static_cast<int> ( d[3] * img.cols ) << " ";
std::cout << static_cast<int> ( d[4] * img.rows ) << " ";
std::cout << static_cast<int> ( d[5] * img.cols ) << " ";
std::cout << static_cast<int> ( d[6] * img.rows ) << std::endl;
int x = int ( d[3] * img.cols );
int y = int ( d[4] * img.rows );
int w = int ( d[5] * img.cols ) - x + 1;
int h = int ( d[6] * img.cols ) - y + 1;
cv::Rect rt ( x, y, w, h );
cv::Scalar clr = cv::Scalar ( 0, 255, 0 );
cv::rectangle ( img, rt, clr );
cv::putText ( img, std::to_string ( d[1] ), cv::Point ( x-10,y-10 ), cv::FONT_HERSHEY_SIMPLEX, 1, clr, 2 );
}
}
imwrite (out_file,img);
return 0;
}