Optimum Pipeline for Visual Terrain Classification Using Improved Bag of Visual Words and Fusion Methods

Figure 7

The training time (training set: 1200 terrain images) for different encoding methods with varying codebook size on the DS1 dataset for SIFT and DSIFT descriptors. (a) SIFT (sparse descriptor); (b) DSIFT (dense descriptor). The numbers in -axis outside (inside) the parenthesis indicate the codebook size for activation (difference)-based coding methods.

(a)

(b)