An End-to-End Robotic Visual Localization Algorithm Based on Deep Learning

Table 1

Details of the overall location network structure.

Layers	Output size	Layers
Conv	112 × 112	7 × 7, 64(3/s2)
Pool	56 × 56	3 × 3 max(s2)
Res1	56 × 56
Res2	28 × 28
Res3	14 × 14
Res4	7 × 7
Dense1	1,024	Fully connected
LSTM	128|4	Long short-term memory networks
Dense2(x)	3	Fully connected
Dense3(q)	4	Fully connected