PnP, or Poll and Pool, is sampling module extension for DETR-type architectures that adaptively allocates its computation spatially to be more efficient. Concretely, the PnP module abstracts the image feature map into fine foreground object feature vectors and a small number of coarse background contextual feature vectors. The transformer models information interaction within the fine-coarse feature space and translates the features into the detection result.
Source: PnP-DETR: Towards Efficient Visual Analysis with TransformersPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Denoising | 22 | 15.94% |
Pose Estimation | 16 | 11.59% |
Image Reconstruction | 8 | 5.80% |
Deblurring | 7 | 5.07% |
Image Restoration | 7 | 5.07% |
Super-Resolution | 6 | 4.35% |
6D Pose Estimation using RGB | 4 | 2.90% |
Object Detection | 4 | 2.90% |
MRI Reconstruction | 3 | 2.17% |
Component | Type |
|
---|---|---|
🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |