Atrous Spatial Pyramid Pooling
Introduced by Chen et al. in DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFsAtrous Spatial Pyramid Pooling (ASSP) is a semantic segmentation module for resampling a given feature layer at multiple rates prior to convolution. This amounts to probing the original image with multiple filters that have complementary effective fields of view, thus capturing objects as well as useful image context at multiple scales. Rather than actually resampling features, the mapping is implemented using multiple parallel atrous convolutional layers with different sampling rates.
Source:
Latest Papers
Tasks
| TASK | PAPERS | SHARE |
|---|---|---|
| Semantic Segmentation | 21 | 39.62% |
| Image Classification | 3 | 5.66% |
| Object Detection | 3 | 5.66% |
| Instance Segmentation | 2 | 3.77% |
| Autonomous Driving | 2 | 3.77% |
| Lake Ice Monitoring | 1 | 1.89% |
| Multi-Task Learning | 1 | 1.89% |
| Pose Estimation | 1 | 1.89% |
| Bird View Synthesis | 1 | 1.89% |
Usage Over Time
Components
| COMPONENT | TYPE |
|
|---|---|---|
Dilated Convolution
|
Convolutions | |
Spatial Pyramid Pooling
|
Pooling Operations |