Learning Condition-Invariant Scene Representations for Place Recognition across the Seasons Using Auto-Encoder and ICA

Table 3

Details of the dataset.

Feature	Value
Total number of images	143072
Total number of training images	128796
Total number of test images	14276
Dimension (pixels)	32 × 64
Color	Grayscale
Total number of images from each season	35768
Total number of training images from each season	32199
Total number of test images from each season	3569