Open manmax31 opened 9 years ago
RootSIFT with kNN + euclidean distance gives accuracy of 75%
SIFT: Logistic 42% LDA 46% Decision Tree Classifier 36% Random Forest 46% AdaBoost with SVM 6% SVM Poly 70% Gaussian NB 26% KNN 78%-84% GradientBoost 48%
HOG: kNN - 40%
DAISY: kNN - 22%
GIST: kNN - 60 - 82% RandomForest - 44% Neural Network - 90% Large DataSet 1 Hidden Layer : 96% DNN: 97.12%
SVM with Polynomial kernel gave the highest score till now with over 80% compared to RBF and LINEAR kernel. Tried RootSIFT instead of SIFT. GradientBoosting Classifier had a score of 63%. AdaBoost with kNN had a low score as well