PaperICCV · 2021
Vision Transformers for Dense Prediction
1page built on it
6more pages cite it
3narratives
0registry papers cite it
In the Atlas
ModelPrimary source of
DPT (Dense Prediction Transformer)
ViT encoder whose tokens from four layers are reassembled into image-like feature maps at multiple resolutions and fused by a convolutional decoder into full-resolution dense predictions; trained for monocular depth with the MiDaS protocol and for semantic segmentation.
Cited as a reference by 6 pages, grouped by domain
Appears in 3 narratives
Lineage
Papers in the Atlas registry only · one dot per paperBuilt upon by · 0
Builds on · 9
2015Fully Convolutional Networks for Semantic SegmentationLong, Shelhamer, Darrell2015U-Net: Convolutional Networks for Biomedical Image SegmentationRonneberger, Fischer, Brox2016Deep Residual Learning for Image RecognitionHe, Zhang et al.2018DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFsChen, Papandreou et al.2019Towards Robust Monocular Depth Estimation: Mixing Datasets for Zero-shot Cross-dataset TransferRanftl, Lasinger et al.2019Deep High-Resolution Representation Learning for Human Pose EstimationSun, Xiao et al.2020An Image is Worth 16x16 Words: Transformers for Image Recognition at ScaleDosovitskiy, Beyer et al.Built upon by · 0