mark/small-image-descriptors
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

WIP on new init

Browse Source 
Rework seems done, but man are there bugs
This commit is contained in:
Mark Hoekveen
2018-10-11 17:05:41 +02:00
parent d0cbf6b420
commit edfa53a4ff
3 changed files with 48 additions and 189 deletions
Download Patch File Download Diff File Expand all files Collapse all files
DescriptorManager.cpp
+36 -181
View File
@@ -83,9 +83,15 @@ void DescriptorManager::Init() {
if (!fs::is_directory(mutationBase)) if (!fs::is_directory(mutationBase))
fs::create_directory(mutationBase); fs::create_directory(mutationBase);
fs::path mutPath(mutationBase / itr->path().filename().generic_string()); fs::path mutPath(mutationBase / itr->path().filename().generic_string());
Image* mutated;
Image* mutated = mutators[i]->getMutated(image); if (!fs::exists(mutPath)) { //only mutate if nothing on disk
mutated->write(mutPath.generic_string()); mutated = mutators[i]->getMutated(image);
mutated->write(mutPath.generic_string());
}
else {
mutated = new Image;
mutated->read(mutPath.generic_string());
}
//now calculate all the features for the mutated images as well //now calculate all the features for the mutated images as well
//foreach descriptor //foreach descriptor
@@ -103,7 +109,7 @@ void DescriptorManager::Init() {
//descriptors[i][image_name] = b1; //descriptors[i][image_name] = b1;
b1 = &descriptor_types[j]->feature(mutated); b1 = &descriptor_types[j]->feature(mutated);
unsigned long long n = b1->to_ullong(); unsigned long long n = b1->to_ullong();
cout << "Writing " << b1 << "(" << n << ")" << " to " << featurePath << endl; cout << "Writing " << *b1 << "(" << n << ")" << " to " << featurePath << endl;
ofs.open(featurePath, fs::fstream::binary); //in binary mode; only open file after calculations have been done ofs.open(featurePath, fs::fstream::binary); //in binary mode; only open file after calculations have been done
ofs.write(reinterpret_cast<const char*>(&n), sizeof(n)); ofs.write(reinterpret_cast<const char*>(&n), sizeof(n));
ofs.close(); ofs.close();
@@ -115,201 +121,52 @@ void DescriptorManager::Init() {
ifs.close(); ifs.close();
b1 = new bitset<64>(n); b1 = new bitset<64>(n);
} }
//Else...?
//Do we need to store mutated features in a cache? For large experiments it's probably faster.
mutatedFeatures[j][i][image_name] = *b1; mutatedFeatures[j][i][image_name] = *b1;
} }
delete mutated; delete mutated;
} }
delete image; delete image;
} }
initialized = true;
} }
//Sorts all images we have on distance from the query image
void DescriptorManager::rankImages(string image_name, int method) {
//Loads all images and determines the feature vectors
bool DescriptorManager::loadDescriptors() {
if (!fs::exists(img_path)) return false;
fs::directory_iterator end_itr; //last file in dir
map<string, Image*> images; //image cache
fs::ifstream ifs;
fs::ofstream ofs;
for (int i = 0; i < numDescriptors; i++) {
//cout << "waddup " << i << endl;
//folder to store descriptors; create if not exists
fs::path descriptorBase("meta" / descriptor_types[i]->ToPath());
if (!fs::is_directory(descriptorBase))
if (fs::create_directory(descriptorBase)) {}
//cout << descriptorBase << " created." << endl;
//foreach image
for (fs::directory_iterator itr(img_path); itr != end_itr; ++itr) {
if (!itr->path().has_extension()) continue;
fs::path featurePath(descriptorBase / itr->path().stem());
//string image_name = featurePath.filename().generic_string();
string image_name = itr->path().filename().generic_string();
if (fs::exists(featurePath)) { //feature already calculated
ifs.open(featurePath, fs::fstream::binary); //in binary mode
unsigned long long n;
ifs.read(reinterpret_cast<char*>(&n), sizeof(n));
ifs.close();
bitset<64> b1(n);
features[i][image_name] = b1;
}
else { //feature not yet calculated
Image* image; //only load image from disk if not loaded before
if (images.count(image_name)) { //if image exists in map
image = images[image_name];
}
else {
image = new Image;
image->read(itr->path().string());
images[image_name] = image;
}
//cout << "Calculating " << image_name << endl;
bitset<64> b1 = descriptor_types[i]->feature(image);
//descriptor_types[i]->distance(b1, b1);
features[i][image_name] = b1;
unsigned long long n = b1.to_ullong();
ofs.open(featurePath, fs::fstream::binary); //in binary mode; only open file after calculations have been done
ofs.write(reinterpret_cast<const char*>(&n), sizeof(n));
ofs.close();
}
}
}
return true;
}
//Loads all images and makes and stores the mutations
bool DescriptorManager::createMutations() {
if (!fs::exists(img_path)) return false;
fs::directory_iterator end_itr; //last file in dir
map<string, Image*> images; //image cache
fs::ifstream ifs;
fs::ofstream ofs;
//foreach mutation
for (int i = 0; i < numMutations; i++) {
fs::path mutationBase(img_path / mutators[i]->ToPath());
if (!fs::is_directory(mutationBase))
if (fs::create_directory(mutationBase)) {}
//foreach image
int z = 0;
for (fs::directory_iterator itr(img_path); itr != end_itr; ++itr) {
fs::path mutPath(mutationBase / itr->path().filename().generic_string());
//if (z++ > 50) break; //for quick execution
//if (fs::exists(mutPath)) continue; //already exists
if (!itr->path().has_extension()) continue; //skip directories
Image* image = new Image(itr->path().string());
Image* mutated = mutators[i]->getMutated(image);
mutated->write(mutPath.generic_string());
//now calculate all the features for the mutated images as well
//foreach descriptor
for (int j = 0; j < numDescriptors; j++) {
//folder to store descriptors; create if not exists
fs::path descriptorBase("meta" / descriptor_types[j]->ToPath() / mutators[i]->ToPath());
//cout << "Doing mutated descriptors for " << descriptorBase << endl;
if (!fs::is_directory(descriptorBase))
if (fs::create_directory(descriptorBase)) {}
fs::path featurePath(descriptorBase / itr->path().stem());
//cout << featurePath << endl;
string image_name = itr->path().filename().generic_string();
if (!fs::exists(featurePath)) { //feature not yet calculated
//cout << "Calculating " << image_name << endl;
//descriptor_types[i]->distance(b1, b1);
//descriptors[i][image_name] = b1;
bitset<64> b1 = descriptor_types[j]->feature(mutated);
unsigned long long n = b1.to_ullong();
cout << "Writing " << b1 << "(" << n << ")" << " to " << featurePath << endl;
ofs.open(featurePath, fs::fstream::binary); //in binary mode; only open file after calculations have been done
ofs.write(reinterpret_cast<const char*>(&n), sizeof(n));
ofs.close();
}
}
delete mutated;
delete image;
}
}
return true;
}
//essentially insertion sort into a map
void DescriptorManager::similarImages(string image_name, int method, int numImages) {
bitset<64> target; bitset<64> target;
if (features[method].count(image_name)) { //if image exists in map if (features[method].count(image_name) && initialized) { //if we have image
target = features[method][image_name]; target = features[method][image_name];
} }
else { else {
Image* image = new Image; return;
image->read(image_name); //assuming http url }
target = descriptor_types[method]->feature(image); distances.clear(); //start fresh
} //TODO: Add support for uploaded images
//cout << "HEY?!" << endl;
//first we need all distances to this image.. and insert them in a sorted way
//multiset<pair<unsigned long long, string> > distances; //multiset is probably not needed, but who knows..
//distances.clear(); //start fresh
for (map<string, bitset<64>>::iterator it = features[method].begin(); it != features[method].end(); ++it) { for (map<string, bitset<64>>::iterator it = features[method].begin(); it != features[method].end(); ++it) {
//cout << "ho " << endl;
unsigned long long d = descriptor_types[method]->distance(target, it->second); unsigned long long d = descriptor_types[method]->distance(target, it->second);
pair<unsigned long long, string> p(d, it->first); pair<unsigned long long, string> p(d, it->first);
distances.insert(p); //pairs are compared by their first element.. distances.insert(p); //pairs are compared by their first element..
} }
int i = 0;
for (std::multiset<pair<unsigned long long, string>>::iterator it = distances.begin(); it != distances.end() && i < numImages; ++it) {
//cout << "hey " << endl;
if (true || it->second != image_name) {
//cout << "(" << it->first << ", " << it->second << ")" << endl;
i++;
}
}
} }
//inserts the descriptors of the mutated versions of the image. //inserts the descriptors of the mutated versions of the image.
//Assumes the distances array is filled with the normal distances //Assumes the distances array is filled with the normal distances
void DescriptorManager::insertMutations(string image_name, int method) { void DescriptorManager::rankMutations(string image_name, int method) {
multiset<pair<unsigned long long, string> > local_distance;
fs::ifstream ifs;
bitset<64> target; bitset<64> target;
if (features[method].count(image_name)) { //if image exists in map if (features[method].count(image_name) && initialized) { //if we have image
target = features[method][image_name]; target = features[method][image_name];
} }
cout << "Finding hits for " << image_name << " which has feature-number " << target.to_ullong() << endl; else {
for (map<string, bitset<64>>::iterator it = features[method].begin(); it != features[method].end(); ++it) { return;
//cout << "ho " << endl;
unsigned long long d = descriptor_types[method]->distance(target, it->second);
pair<unsigned long long, string> p(d, it->first);
local_distance.insert(p); //pairs are compared by their first element..
} }
cout << "Finding hits for " << image_name << " which has feature-number " << target.to_ullong() << endl;
multiset<pair<unsigned long long, string> > local_distance = distances; //get our local copy of the distances.
for (int i = 0; i < numMutations; i++) { for (int i = 0; i < numMutations; i++) {
fs::path descriptorBase("meta" / descriptor_types[method]->ToPath() / mutators[i]->ToPath()); bitset<64> b1 = mutatedFeatures[method][i][image_name];
fs::path imagePath(image_name); unsigned long long d = descriptor_types[method]->distance(target, b1);
fs::path featurePath(descriptorBase / imagePath.stem()); pair<unsigned long long, string> p(d, mutators[i]->ToString());
cout << "Finding mutation for this image at: " << featurePath << endl; local_distance.insert(p);
if (fs::exists(featurePath)) { //feature already calculated
ifs.open(featurePath, fs::fstream::binary); //in binary mode
unsigned long long n;
ifs.read(reinterpret_cast<char*>(&n), sizeof(n));
ifs.close();
cout << "Number: " << n << endl;
bitset<64> b1(n);
unsigned long long d = descriptor_types[method]->distance(target, b1);
pair<unsigned long long, string> p(d, mutators[i]->ToString());
local_distance.insert(p);
}
else {
cout << "data not set properly" << endl;
}
} }
cout << "..." << endl; cout << "..." << endl;
for (int i = 0; i < numMutations; i++) { for (int i = 0; i < numMutations; i++) {
int j = 0; int j = 0;
@@ -324,19 +181,10 @@ void DescriptorManager::insertMutations(string image_name, int method) {
} }
} }
void DescriptorManager::checkImages(int method) {
fs::directory_iterator end_itr;
for (fs::directory_iterator itr(img_path); itr != end_itr; ++itr) {
if (!itr->path().has_extension()) continue; //skip directories
insertMutations(itr->path().string(), method);
}
}
//outputs the gathered data to iostream //outputs the gathered data to iostream
//can be called directly or is called by its overload //can be called directly or is called by its overload
//assumes a < numDescriptors //assumes a < numDescriptors
void DescriptorManager::outputMap(int a) { void DescriptorManager::outputMap(int a) {
if (a != 2) return;
cout << "Outputting map " << descriptor_types[a]->ToString() << endl; cout << "Outputting map " << descriptor_types[a]->ToString() << endl;
map<string, bitset<64>> a_map = features[a]; map<string, bitset<64>> a_map = features[a];
if (a_map.begin() == a_map.end()) { if (a_map.begin() == a_map.end()) {
@@ -353,4 +201,11 @@ void DescriptorManager::outputMap() {
for (int i = 0; i < numDescriptors; i++) { for (int i = 0; i < numDescriptors; i++) {
outputMap(i); outputMap(i);
} }
}
//Outputs all generated features ordered on distance from the query image.
void DescriptorManager::outputRank(string imagename, int method) {
for (std::multiset<pair<unsigned long long, string>>::iterator it = distances.begin(); it != distances.end(); ++it) {
cout << "(" << it->first << ", " << it->second << ")" << endl;
}
} }
DescriptorManager.h
+5 -5
View File
@@ -27,18 +27,18 @@ public:
DescriptorManager(const fs::path & image_path); DescriptorManager(const fs::path & image_path);
~DescriptorManager(); ~DescriptorManager();
void Init(); void Init();
bool loadDescriptors(); void rankImages(string image_name, int method);
bool createMutations(); void rankMutations(string image_name, int method);
void similarImages(string image_name, int method, int numImages);
void insertMutations(string image_name, int method);
void checkImages(int method);
void outputMap(int a); void outputMap(int a);
void outputMap(); void outputMap();
void DescriptorManager::outputRank(string imagename, int method);
private: private:
static const int numDescriptors = 8; static const int numDescriptors = 8;
static const int numMutations = 7; static const int numMutations = 7;
//TODO: automagically update this? //TODO: automagically update this?
bool initialized = false;
map<string, bitset<64>> features[numDescriptors]; //array of maps that will hold the descriptors map<string, bitset<64>> features[numDescriptors]; //array of maps that will hold the descriptors
map<string, bitset<64>> mutatedFeatures[numDescriptors][numMutations]; map<string, bitset<64>> mutatedFeatures[numDescriptors][numMutations];
multiset<pair<unsigned long long, string> > distances; //multiset that stores the distances multiset<pair<unsigned long long, string> > distances; //multiset that stores the distances
main.cpp
+7 -3
View File
@@ -23,7 +23,7 @@ using namespace Magick;
int main(int argc, char **argv) { int main(int argc, char **argv) {
string imagename = "im124.jpg"; string imagename = "im80.jpg";
int method = 6; int method = 6;
int count = 1000; int count = 1000;
if (argc > 2) { if (argc > 2) {
@@ -41,12 +41,16 @@ int main(int argc, char **argv) {
InitializeMagick(*argv); InitializeMagick(*argv);
DescriptorManager* manager = new DescriptorManager("data"); DescriptorManager* manager = new DescriptorManager("data");
manager->Init(); manager->Init();
manager->rankImages(imagename, method);
manager->rankMutations(imagename, method);
//manager->outputMap(method);
manager->outputRank(imagename, method);
//cout << "Loading descriptiors.."; //cout << "Loading descriptiors..";
//manager->loadDescriptors(); //manager->loadDescriptors();
manager->createMutations(); //manager->createMutations();
//cout << "done?" << endl; //cout << "done?" << endl;
//manager->similarImages(imagename, method, count); //manager->similarImages(imagename, method, count);
manager->insertMutations(imagename, method); //manager->insertMutations(imagename, method);
//cout << "done!" << endl << "Calculating distances.." << endl; //cout << "done!" << endl << "Calculating distances.." << endl;
//cout << "done!" << endl << "Outputting map.." << endl; //cout << "done!" << endl << "Outputting map.." << endl;
//manager->outputMap(); //manager->outputMap();