Skip to content

Latest commit

 

History

History

tinycd

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 

TinyCD

TINYCD: A (Not So) Deep Learning Model For Change Detection

Introduction

Official Repo

Code Snippet

Abstract

In this paper, we present a lightweight and effective change detection model, called TinyCD. This model has been designed to be faster and smaller than current state-of-the-art change detection models due to industrial needs. Despite being from 13 to 140 times smaller than the compared change detection models, and exposing at least a third of the computational complexity, our model outperforms the current state-of-the-art models by at least 1% on both F1 score and IoU on the LEVIR-CD dataset, and more than 8% on the WHU-CD dataset. To reach these results, TinyCD uses a Siamese U-Net architecture exploiting low-level features in a globally temporal and locally spatial way. In addition, it adopts a new strategy to mix features in the space-time domain both to merge the embeddings obtained from the Siamese backbones, and, coupled with an MLP block, it forms a novel space-semantic attention mechanism, the Mix and Attention Mask Block (MAMB). Source code, models and results are available here: this https URL

@article{codegoni2022tinycd,
  title={TINYCD: A (Not So) Deep Learning Model For Change Detection},
  author={Codegoni, Andrea and Lombardi, Gabriele and Ferrari, Alessandro},
  journal={arXiv preprint arXiv:2207.13159},
  year={2022}
}

Results and models

LEVIR-CD

Method Backbone Crop Size Lr schd Mem (GB) Precision Recall F1-Score IoU config download
TinyCD EfficientNet 256x256 40000 91.87 89.89 90.87 83.26 config
  • All metrics are based on the category "change".
  • All scores are computed on the test set.