small-image-descriptors/DescriptorManager.cpp

#include "DescriptorManager.h"

//Default constructor, don't use this
DescriptorManager::DescriptorManager() {
	img_path = "data"; //default value
}

//Sets image path and defines the image descriptors.
DescriptorManager::DescriptorManager(const fs::path & image_path) {
	img_path = image_path;

	//Manual definitions.. SAD!
	/*descriptor_types[7] = new DCTMedian();
	descriptor_types[6] = new DCTPearson();
	descriptor_types[5] = new threshold();
	descriptor_types[4] = new pearson();
	descriptor_types[3] = new sobel();
	descriptor_types[2] = new median();
	descriptor_types[1] = new numColors();*/
	descriptor_types[0] = new averageColor();

	mutators[0] = new Grayscale();
	mutators[1] = new Crop();
	mutators[2] = new Flop();
	mutators[3] = new JPEG();
	mutators[4] = new LowQ();
	mutators[5] = new Hue();
	mutators[6] = new Brighten();
	mutators[7] = new Watermark();
}

//No need to call destructor.. yet?
DescriptorManager::~DescriptorManager() {
	//Do some clean-up?
	//Realistically, when this gets destroyed the application is done anyway.
	//Let the OS take care of garbage collection
}

//Loads all base images, calculates their descriptors,
//calculates all image mutations and stores their descriptors
//on disk.
void DescriptorManager::Init() {
	//First load all images and store descriptors in cache and disk
	if (!fs::exists(img_path)) {
		cout << "Error. Data folder not found." << endl;
		return;
	}
	fs::ifstream ifs;
	fs::ofstream ofs;
	//LoadFeatures();

	numImages = std::count_if(
		fs::directory_iterator(img_path),
		fs::directory_iterator(),
		static_cast<bool(*)(const fs::path&)>(fs::is_regular_file));
	//foreach image in directory (non-recursive)
	int filecount = 0;
	for (fs::directory_iterator itr(img_path); itr != fs::directory_iterator(); ++itr) {
		if (!itr->path().has_extension()) continue; //skip directories and such
		filecount++; 
		string image_name = itr->path().filename().generic_string();
		cout << setw(60) << setfill('*') << "\r" << string(60, ' ') << flush;  
		cout << setfill('*') << setw(60) << right<< "\rProgress: " << filecount*100/numImages << "% - (" << filecount << "/" << numImages << "): " << image_name << flush;
		Image* image = new Image;
		image->read(itr->path().string());
		//foreach descriptor
		for (int i = 0; i < numDescriptors; i++) {
			//First check if we already have this information from the loaded cache
			if (features[i].count(image_name) > 0) {
				//cout << "We already have feature for " << image_name << endl;
				continue;
			}
			//folder to store descriptors; create if not exists
			fs::path descriptorBase("meta" / descriptor_types[i]->ToPath());
			if (!fs::is_directory(descriptorBase))
				fs::create_directory(descriptorBase);
			//cout << "Starting with descriptor " << i << " on location " << descriptorBase.generic_string() << endl;

			fs::path featurePath(descriptorBase / itr->path().stem());
			uint64_t b1;
			if (fs::exists(featurePath)) { //feature on disk, load it in
				ifs.open(featurePath, fs::fstream::binary);
				ifs.read(reinterpret_cast<char*>(&b1), sizeof(b1));
				ifs.close();
			}
			else { //not on disk, calculate it
				b1 = descriptor_types[i]->feature(image);
				ofs.open(featurePath, fs::fstream::binary);
				ofs.write(reinterpret_cast<const char*>(&b1), sizeof(b1));
				ofs.close();
			}
			features[i][image_name] = b1;
			//SaveFeatures();
		}
		for (int i = 0; i < numMutations; i++) {
			//Check if we have somewhere to store the mutation
			fs::path mutationBase(img_path / mutators[i]->ToPath());
			if (!fs::is_directory(mutationBase))
				fs::create_directory(mutationBase);
			bool missing = false;
			for (int j = 0; j < numDescriptors; j++) {
				if (mutatedFeatures[j][i].count(image_name) <= 0) {
					missing = true;
				}
			}
			fs::path mutPath(mutationBase / itr->path().filename().generic_string());
			Image* mutated;
			if (!fs::exists(mutPath)) { //only mutate if nothing on disk
				//Measured: about 10ms
				auto start = high_resolution_clock::now(); 
				mutated = mutators[i]->getMutated(image);
				mutated->write(mutPath.generic_string());
				auto duration = duration_cast<microseconds>(high_resolution_clock::now() - start); 
				//cout << "Calculating mutation " << mutators[i]->ToString() << " took " << duration.count() << "μs." << endl;
				
			}
			else { 
				//Measured: about 4ms
				auto start = high_resolution_clock::now();
				mutated = new Image;
				mutated->read(mutPath.generic_string());
				auto duration = duration_cast<microseconds>(high_resolution_clock::now() - start); 
				//cout << "Loading mutation " << mutators[i]->ToString() << " took " << duration.count() << "μs." << endl;
			}

			//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());
				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();
				uint64_t b1;
				if (!fs::exists(featurePath)) { //feature not yet calculated
												//descriptor_types[i]->distance(b1, b1);
												//descriptors[i][image_name] = b1;
					b1 = descriptor_types[j]->feature(mutated);
					//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*>(&b1), sizeof(b1));
					ofs.close();
				}
				else {
					ifs.open(featurePath, fs::fstream::binary);
					ifs.read(reinterpret_cast<char*>(&b1), sizeof(b1));
					ifs.close();
				}

				mutatedFeatures[j][i][image_name] = b1;
			}
			delete mutated;
		}
		delete image;
	}
	initialized = true;
}


//Saves feature map to disk.
//Makes it so that it can reload easily
void DescriptorManager::SaveFeatures() {
	{ //Boost serialization
		ofstream f("features.map", ios::binary);
		//ofstream feature_cache_backup("features.map.bak", ios::binary);
		//feature_cache_backup << feature_cache.rdbuf();
		ar::binary_oarchive oa(f);
		oa << features;

		ofstream fm("mutated_features.map", ios::binary);
		ar::binary_oarchive oam(fm);
		oam << mutatedFeatures;
	}
	/*{ //YAS serialization
		yas::file_ostream os("features.yas", yas::file_mode::file_trunc);
		//const std::size_t flg = yas::binary|yas::file;
		yas::binary_oarchive<yas::file_ostream, yas::file|yas::binary> oa(os);
		oa & features;
		//yas::save<yas::file|yas::binary>(image_name, features);
	}*/
}

//Loads saved feature maps from disk.
void DescriptorManager::LoadFeatures() {
	{ //Boost
		ifstream f("features.map", ios::binary);
		if (f.is_open()) {
			ar::binary_iarchive bi(f);
			bi >> features;
		}
		ifstream fm("mutated_features.map", ios::binary);
		if (fm.is_open()) {
			ar::binary_iarchive bim(fm);
			bim >> mutatedFeatures;
		}
	}
	{ //YAS 

	}
}

//Sorts all base (unmutated) images in order of distance from the given image
//After this function the distances array will be filled
void DescriptorManager::rankImages(string image_name, int method) {
	uint64_t target;
	if (features[method].count(image_name) && initialized) { //if we have image
		target = features[method][image_name];
	}
	else {
		return;
	}
	distances.clear(); //start fresh
	for (map<string, uint64_t>::iterator it = features[method].begin(); it != features[method].end(); ++it) {
		unsigned long long d = descriptor_types[method]->distance(target, it->second);
		pair<unsigned long long, string> p(d, it->first);
		distances.insert(p); //pairs are compared by their first element..
	}
}

//Adds mutated descriptors of a given image to a copy of the distances array.
//It inserts these mutated descriptors in a sorted way
//Assumes the distances array is filled with the normal distances
multiset<pair<unsigned long long, string>> DescriptorManager::rankMutations(string image_name, int method) {
	uint64_t target;
	if (features[method].count(image_name) && initialized) { //this should always be the case
		target = features[method][image_name];
	}

	//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++) {
		uint64_t b1 = mutatedFeatures[method][i][image_name];
		unsigned long long d = descriptor_types[method]->distance(target, b1);
		pair<unsigned long long, string> p(d, mutators[i]->ToString());
		local_distance.insert(p);
		//upper_bound
	}
	return local_distance;
	/*
	//cout << "..." << endl;
	for (int i = 0; i < numMutations; i++) {
		int j = 0;
		for (std::multiset<pair<unsigned long long, string>>::iterator it = local_distance.begin(); it != local_distance.end(); ++it) {
			//cout << "hey " << endl;
			if (it->second == mutators[i]->ToString()) {
				cout << j << "th position: ";
				cout << "(" << it->first << ", " << it->second << ")" << endl;
			}
			j++;
		}
	}*/
}

//Needs to be called after Init().
//Takes all the calculated features and rankes 
void DescriptorManager::runExperiments() {
	fs::ofstream ofs;
	for (int i = 0; i < numDescriptors; i++) {
		for (map<string, uint64_t>::iterator it = features[i].begin(); it != features[i].end(); ++it) {
			//cout << "=== " << descriptor_types[i]->ToString() << ": " << it->first << " ===" << endl;
			rankImages(it->first, i);
			multiset<pair<unsigned long long, string>> local_distance = rankMutations(it->first, i);

			//Now we can draw some conclusions from the received data:
			//Put all results in some useful files in the results directory:
			fs::path resultFolder("results" / descriptor_types[i]->ToPath());
			if (!fs::is_directory(resultFolder)) //TODO: Create all folders in Init()
				fs::create_directory(resultFolder);
			for (int i = 0; i < numMutations; i++) {
				int j = 0;
				fs::path resultPath(resultFolder / mutators[i]->ToPath());
				ofs.open(resultPath, ofstream::out | ofstream::app);
				for (std::multiset<pair<unsigned long long, string>>::iterator it = local_distance.begin(); it != local_distance.end(); ++it) {
					if (it->second == mutators[i]->ToString()) {
						double percentile = (j + 1) / (double)(numImages+numMutations);

						ofs << (int)(percentile*100) << endl;
						//cout << "(" << it->first << ", " << it->second << ")" << endl;
					}
					j++;
				}
				ofs.close();
			}

		}
	}
}

//outputs the gathered data to iostream
//can be called directly or is called by its overload
//assumes a < numDescriptors
void DescriptorManager::outputMap(int a) {
	cout << "Outputting map " << descriptor_types[a]->ToString() << endl;
	map<string, uint64_t> a_map = features[a];
	if (a_map.begin() == a_map.end()) {
		cout << "------------------------" << endl << "No data found" << endl << endl;
	}

	for (map<string, uint64_t>::iterator it = a_map.begin(); it != a_map.end(); ++it) {
		cout << it->first << " => " << it->second << " (" << bitset<64>(it->second) << ")" << endl;
	}
} 

//calls its buddy function to output all data to iostream
void DescriptorManager::outputMap() {
	for (int i = 0; i < numDescriptors; 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;
	}
}

//Just runs the preview functions of all the descriptors on given image.
//No need to init before this.
void DescriptorManager::generatePreviews(string imagename) {
	fs::path imagePath(img_path / imagename);
	Image* img = new Image(imagePath.generic_string());
	for (int i = 0; i < numDescriptors; i++) {
		descriptor_types[i]->preview(img);
	}
}