Skip to main content
. 2023 Jun 9;36(5):2194–2209. doi: 10.1007/s10278-023-00832-x

Table 2.

Deep feature size Inception V3 for each selected layer

Feature size
Layer Output shape Max/Avg flatten
mixed0 35x35x256 256 313600
mixed1 35x35x288 288 352800
mixed2 35x35x288 288 352800
mixed3 17x17x768 768 221952
mixed4 17x17x768 768 221952
mixed5 17x17x768 768 221952
mixed6 17x17x768 768 221952
mixed7 17x17x768 768 221952
mixed8 8x8x1280 1280 81920
mixed9 8x8x2048 2048 131072
mixed10 8x8x2048 2048 131072