Fully Transformer Networks for Semantic Image Segmentation
pith:HAINFX3Mopen to challenge →
read the original abstract
Transformers have shown impressive performance in various natural language processing and computer vision tasks, due to the capability of modeling long-range dependencies. Recent progress has demonstrated that combining such Transformers with CNN-based semantic image segmentation models is very promising. However, it is not well studied yet on how well a pure Transformer based approach can achieve for image segmentation. In this work, we explore a novel framework for semantic image segmentation, which is encoder-decoder based Fully Transformer Networks (FTN). Specifically, we first propose a Pyramid Group Transformer (PGT) as the encoder for progressively learning hierarchical features, meanwhile reducing the computation complexity of the standard Visual Transformer (ViT). Then, we propose a Feature Pyramid Transformer (FPT) to fuse semantic-level and spatial-level information from multiple levels of the PGT encoder for semantic image segmentation. Surprisingly, this simple baseline can achieve better results on multiple challenging semantic segmentation and face parsing benchmarks, including PASCAL Context, ADE20K, COCOStuff, and CelebAMask-HQ. The source code will be released on https://github.com/BR-IDL/PaddleViT.
This paper has not been read by Pith yet.
Forward citations
Cited by 4 Pith papers
-
VesselRW: Weakly Supervised Subcutaneous Vessel Segmentation via Learned Random Walk Propagation
VesselRW expands sparse vessel annotations into dense probabilistic supervision via a jointly trained differentiable random walk model with uncertainty weighting and topology regularization for CNN-based subcutaneous ...
-
DualResolution Residual Architecture with Artifact Suppression for Melanocytic Lesion Segmentation
Dual-resolution residual architecture with boundary-aware connections, channel attention, artifact suppression, and combined Dice-Tversky plus boundary and contrastive losses improves lesion boundary precision over st...
-
Edge Detection for Organ Boundaries via Top Down Refinement and SubPixel Upsampling
A top-down backward refinement network with subpixel upsampling generates crisp high-resolution organ boundaries in medical images and improves downstream segmentation and registration performance.
-
Deeply Dual Supervised learning for melanoma recognition
A dual-pathway deep learning model with attention mechanisms and multi-scale feature aggregation claims superior accuracy and fewer false positives for melanoma detection on benchmark datasets.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.