DropBlock is a structured form of dropout directed at regularizing convolutional networks. In DropBlock, units in a contiguous region of a feature map are dropped together. As DropBlock discards features in a correlated area, the networks must look elsewhere for evidence to fit the data.
Source: DropBlock: A regularization method for convolutional networksPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Object Detection | 58 | 26.73% |
Real-Time Object Detection | 10 | 4.61% |
Image Classification | 9 | 4.15% |
Person Re-Identification | 8 | 3.69% |
Semantic Segmentation | 7 | 3.23% |
Autonomous Driving | 6 | 2.76% |
General Classification | 5 | 2.30% |
Classification | 4 | 1.84% |
Instance Segmentation | 4 | 1.84% |
Component | Type |
|
---|---|---|
🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |