Image Model Blocks

Inception Module

Introduced by Szegedy et al. in Going Deeper with Convolutions

An Inception Module is an image model block that aims to approximate an optimal local sparse structure in a CNN. Put simply, it allows for us to use multiple types of filter size, instead of being restricted to a single filter size, in a single image block, which we then concatenate and pass onto the next layer.

Source: Going Deeper with Convolutions


Paper Code Results Date Stars


Task Papers Share
Image Classification 31 12.16%
General Classification 30 11.76%
Classification 21 8.24%
Object Detection 18 7.06%
Quantization 13 5.10%
Object Recognition 8 3.14%
Autonomous Driving 5 1.96%
Domain Adaptation 5 1.96%
Image Generation 4 1.57%