[CVPR 2022--Oral] Restormer: Efficient Transformer for High-Resolution Image Restoration. SOTA for motion deblurring, image deraining, denoising (Gaussian/real data), and defocus deblurring.
-
Updated
Oct 23, 2025 - Python
[CVPR 2022--Oral] Restormer: Efficient Transformer for High-Resolution Image Restoration. SOTA for motion deblurring, image deraining, denoising (Gaussian/real data), and defocus deblurring.
Mask Transfiner for High-Quality Instance Segmentation, CVPR 2022
Implementation of the Transformer variant proposed in "Transformer Quality in Linear Time"
Official PyTorch Implementation of Long-Short Transformer (NeurIPS 2021).
[CVPR 2023] IMP: iterative matching and pose estimation with transformer-based recurrent module
[MICCAI 2023] DAE-Former: Dual Attention-guided Efficient Transformer for Medical Image Segmentation
[NeurIPS'21] "Chasing Sparsity in Vision Transformers: An End-to-End Exploration" by Tianlong Chen, Yu Cheng, Zhe Gan, Lu Yuan, Lei Zhang, Zhangyang Wang
[NeurIPS 2022 Spotlight] This is the official PyTorch implementation of "EcoFormer: Energy-Saving Attention with Linear Complexity"
Official PyTorch implementation of our ECCV 2022 paper "Sliced Recursive Transformer"
[ICLR 2022] "Unified Vision Transformer Compression" by Shixing Yu*, Tianlong Chen*, Jiayi Shen, Huan Yuan, Jianchao Tan, Sen Yang, Ji Liu, Zhangyang Wang
This repo is a alternative solution for standard self-attention. It uses tree routing with learnable functions for cheaper self-attention while not compromising accuracy.
[ICCV 2023] Efficient Video Action Detection with Token Dropout and Context Refinement
Official Implementation of Energy Transformer in PyTorch for Mask Image Reconstruction
This repository contains the official code for Energy Transformer---an efficient Energy-based Transformer variant for graph classification
Official Repository of "Fibottention: Inceptive Visual Representation Learning with Diverse Attention Across Heads"
Demo code for CVPR2023 paper "Sparsifiner: Learning Sparse Instance-Dependent Attention for Efficient Vision Transformers"
Temporalmesh-transformer. It is the first architecture to simultaneously fuse dynamic graph topology, token-level adaptive compute, and temporal semantic decay into a single unified model. No prior work does all three together.
Two Percent Attention: Benchmarking Sparse Attention in Vision Transformers
Gated Attention Unit (TensorFlow implementation)
Full-context associative memory via compact equilibrium solves, with linear sequence scaling and PyTorch/Triton acceleration.
To associate your repository with the efficient-transformers topic, visit your repo's landing page and select "manage topics."