search

kobiso / SENet-tensorflow-slim

SENet implementation on TensorFlow Slim
MIT License
58 stars 26 forks source link

SE-net module placement in inception resnet #2

Closed sp-ananth closed 6 years ago

sp-ananth commented 6 years ago

Hi there, I notice that the original paper says that for resnet, the SE-block is applied to "The non-identity branch of the residual nodule". And in the inception network after each inception block.

In you inception-resnet implementation, you seem to have applied the SE module to the identity branch of each block instead. (i.e) to the net variable rather than the (up*scale) variable.

Is there any reason for this? Did this configuration work for you?

kobiso commented 6 years ago

@sp-ananth I just checked my code and what you said is right. I meant to put SE module on the non-identity branch of the residual module, but I mistakenly put SE module to the identity branch. Thank you for pointing out my mistake 👍 I will revised the code and upload it soon.

kobiso commented 6 years ago

Revising the SE module placement in Inception_resnet_v2 is completed: https://github.com/kobiso/SENet-tensorflow-slim/commit/51b6d27b57498ce163d3b21acdb45292d1857619