02105nam a22001697a 4500999001700000082002700017100007100044245006000115260003000175300000800205500141800213700006001631856003201691942001101723952010001734952010101834  c64812d64809  aR/IMS-19a15-MS(IT)-14  aDharejo, Hina a15-MS(IT)-14aSupervisor - Dr. Fareed Ahmed Jokhio  aAnalysis of Feature Extraction and Image Classification  aNawabshah:bQuest,c2019.  a56p  aABTRACT

In this thesis performance  a classifier i.e. Support machine (SVM) linear kernel mapping is analyzed. various	classes of images of   Dataset   PASCAL VOC 2007 are used for classification. The dataset of consists of different classes whose features are extracted  by  using  Scale Invariant	Feature transform (SIFT)  and quantized into visual words. Then  he frequency of  vectors is recorded in a histogram for each spatial tile of the image. Then the resulting  vectors (features	are used to train the (SVM) for different class of images and  evaluate the  performance of the classifier. The analyzing results give us the performance accuracy of different classes of images with different representation and performance of	the SVM classifier is measured by using different parameters such ascomputation Time: Accuracy. Precision and Recall curveand  Average Precision.
SVM is powerful classification algorithm alternative to neural network. A.s compare to the neural network SVM classifier gives high generalization performance when the measurement of the input pace is very high and no need to add a prior knowledge.

It is observed from theparameters  PR Curve and AP value that the performance of the classifier is increased a it learn m re and m r positive images given are contained in the dataset.

Keywords: feature Extraction. Image Classification. Linear kernel. SIIFT Support Vector machine.
  aDepartment of information technology faculty of science  uhttp://tinyurl.com/v6wxdwnr  cTHESIS  00104070aRESEARCHbRESEARCHd2019-07-01l0oR/IMS-19pMP/41-447r2019-07-01 00:00:00yTHESIS  00104070aRESEARCHbRESEARCHd2023-12-15l0pMP/54682r2023-12-15 00:00:00w2023-12-15yTHESIS