danielgoncalvesti / image-classification Goto Github PK

This project forked from yuhanzha/image-classification

In this project, we improved a classification baseline model for images of dogs, fried chickens and blueberry muffins in terms of ruing time cost and prediction accuracy. The baseline model for comparison uses Gradient Boosting Machine (GBM) with decision stumps on 2000 SIFT features. Our group uses 3 image feature selection models, including SIFT, RGB, LBP and the arbitrary combinations of them, and 8 classification models, including GBM, SVM, XgBoost, Random Forest, Neural Network, AdaBoost, Logistic Regression and Classification Trees.