Vision Transformers for Dense Prediction (ICCV 2021)

Vision Transformers for Dense Prediction (ICCV 2021) - State f the art Real-time (30 FPS and higher) neural network for Semantic segmentation and Mono-Depth estimation from one RGB image. DPT (Dense Prediction Transformers): github.com/int...
René Ranftl, Alexey Bochkovskiy, Vladlen Koltun
openaccess.the...