Convolutions
Spatially Separable Convolution
A Spatially Separable Convolution decomposes a convolution into two separate operations. In regular convolution, if we have a 3 x 3 kernel then we directly convolve this with the image. We can divide a 3 x 3 kernel into a 3 x 1 kernel and a 1 x 3 kernel. Then, in spatially separable convolution, we first convolve the 3 x 1 kernel then the 1 x 3 kernel. This requires 6 instead of 9 parameters compared to regular convolution, and so it is more parameter efficient (additionally less matrix multiplications are required).
Image Source: Kunlun Bai
Papers
| Paper | Code | Results | Date | Stars |
|---|
Tasks
| Task | Papers | Share |
|---|---|---|
| Image Classification | 4 | 17.39% |
| Object Detection | 4 | 17.39% |
| Keypoint Detection | 2 | 8.70% |
| Semantic Segmentation | 2 | 8.70% |
| Audio Fingerprint | 1 | 4.35% |
| Music Information Retrieval | 1 | 4.35% |
| Retrieval | 1 | 4.35% |
| Image Augmentation | 1 | 4.35% |
| Quantization | 1 | 4.35% |
Usage Over Time
Components
| Component | Type |
|
|---|---|---|
| 🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |
