Skip to content

Latest commit

 

History

6 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DualComp

Research code for “Semantic-Geometric Dual Compression: Training-Free Visual Token Reduction for Ultra-High-Resolution Remote Sensing Understanding.”

Results in this repository are transcribed from the manuscript and are not independently reproduced.

Overview

DualComp routes a visual-token budget between semantic and geometric streams:

  • Router predicts task-adaptive lambda and rho controls.
  • SCSA aggregates spatially contiguous semantic regions.
  • IGSR retains geometry-sensitive anchors and completes structural paths.
  • Fusion concatenates the two streams using lambda-aware scaling.

The repository provides a paper-bounded reference implementation. Parameters omitted by the manuscript remain explicit, and no end-to-end host-model adapter is claimed.

Install

python -m venv .venv
source .venv/bin/activate
python -m pip install -e .

Python 3.10 and 3.11 are supported.

Minimal example

import torch
from dualcomp import lambda_weighted_concatenation

semantic = torch.ones(1, 2, 4)
geometric = torch.full((1, 1, 4), 2.0)
output = lambda_weighted_concatenation(semantic, geometric, duality=0.25)

assert output.shape == (1, 3, 4)

Paper-reported results

Item Reported value
XLRS-Bench average 53.1%
Compression ratio 42.4x
Inference speed 3.87 s/image
Qwen2.5-VL-7B transfer 47.9% at 10.24x

These values are copied from the manuscript and are not independent reruns.

Citation and license

Citation metadata is in CITATION.cff. Code in this repository is released under Apache-2.0; external assets remain subject to their own terms.

About

Official code for Semantic-Geometric Dual Compression (DualComp)

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages