NAS-FPN is a Feature Pyramid Network that is discovered via Neural Architecture Search in a novel scalable search space covering all cross-scale connections. The discovered architecture consists of a combination of top-down and bottom-up connections to fuse features across scales
Source: NAS-FPN: Learning Scalable Feature Pyramid Architecture for Object DetectionPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Object Detection | 9 | 34.62% |
Semantic Segmentation | 4 | 15.38% |
Instance Segmentation | 3 | 11.54% |
Image Augmentation | 2 | 7.69% |
General Classification | 2 | 7.69% |
Image Classification | 2 | 7.69% |
Real-Time Object Detection | 2 | 7.69% |
Point Cloud Registration | 1 | 3.85% |
Robust Object Detection | 1 | 3.85% |
Component | Type |
|
---|---|---|
Batch Normalization
|
Normalization | |
Convolution
|
Convolutions | |
Global Average Pooling
|
Pooling Operations | |
Neural Architecture Search
|
Neural Architecture Search | |
ReLU
|
Activation Functions |