A Selective Kernel Convolution is a convolution that enables neurons to adaptively adjust their RF sizes among multiple kernels with different kernel sizes. Specifically, the SK convolution has three operators – Split, Fuse and Select. Multiple branches with different kernel sizes are fused using softmax attention that is guided by the information in these branches. Different attentions on these branches yield different sizes of the effective receptive fields of neurons in the fusion layer
Source: Selective Kernel NetworksPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Object Detection | 4 | 10.00% |
Semantic Segmentation | 4 | 10.00% |
Image Classification | 3 | 7.50% |
Point Cloud Completion | 2 | 5.00% |
Denoising | 2 | 5.00% |
Image Denoising | 2 | 5.00% |
Image Segmentation | 1 | 2.50% |
3D Classification | 1 | 2.50% |
Classification | 1 | 2.50% |
Component | Type |
|
---|---|---|
Batch Normalization
|
Normalization | |
Channel-wise Soft Attention
|
Attention Mechanisms | |
Dilated Convolution
|
Convolutions | |
ReLU
|
Activation Functions | |
Softmax
|
Output Functions |