Convolutions

Pointwise Convolution

Pointwise Convolution is a type of convolution that uses a 1x1 kernel: a kernel that iterates through every single point. This kernel has a depth of however many channels the input image has. It can be used in conjunction with depthwise convolutions to produce an efficient class of convolutions known as depthwise-separable convolutions.

Image Credit: Chi-Feng Wang

Papers


Paper Code Results Date Stars

Tasks


Task Papers Share
Image Classification 70 10.51%
Object Detection 49 7.36%
Classification 37 5.56%
Semantic Segmentation 32 4.80%
Quantization 30 4.50%
Instance Segmentation 11 1.65%
Management 9 1.35%
Ensemble Learning 8 1.20%
Computational Efficiency 8 1.20%

Components


Component Type
🤖 No Components Found You can add them if they exist; e.g. Mask R-CNN uses RoIAlign

Categories