This project is the detector of specific objects on images (e.g. airplanes). Completed as a part of MSU University course of Computer graphics. Main file of the program is src/task2.cpp, which contains the main program that can be executed in two modes: binary classifier and multi-class classifier.
Program performs several steps on the training stage and the prediction stage. Training stage:
- loads images of the training set (data/binary/train or data/multiclass/train) with their labels (data/binary/train_labels.txt or data/multiclass/train_labels.txt)
- extracts features
- trains SVM classifier
- prints model in a file
Testing stage:
- loads images of the test set
- prints results based on the prepared model
Script compare.py is provided which can measure accuracy of completed classification.
To compile the program, run in the command shell from the project's root directory
make all
To pass through training and testing stage, go to build/bin/ and run
task2.exe -d ../../data/binary/train_labels.txt -m model.txt --train
task2.exe -d ../../data/binary/test_labels.txt -m model.txt -l predictions.txt --predict
To remove compiled executables, run in the command shell from the project's root directory
make clean