one vs one is implemented for multiclass classification on the datasets of images containing cats,dogs,horses,humans. SVM Classifiers using Pegasos algorithm is used and nC2 Classifiers are Trained Datasets for horse cat human and dog is used there are total 808 images(202 each) nC2 CLassifiers are trained kernels are used for accuracy RBF kernel gives best accuracy