After this, our task is to annotate the matched object in
the query image. The images in our database are prior annotated
according to the scene or object present. On nding
the best match to the query, our task is to transfer the corresponding
annotation to the query image. The entire process
when performed on a typical desktop, needs considerable
amount of main memory(RAM), storage and processing
power. Apart from storing the images, The RANSAC based
spatial verication of two images uses the 128-dimensional
SIFT vectors and this also needs to be stored.