GenAI4Urban-Energy

May 21, 2026 · View on GitHub

Introduction

The code implementation for the paper: SENSE: Satellite-based ENergy Synthesis for Sustainable Environment., which is accepted by Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2026).(Oral, Top 6% in AI4S Track).

Installation

Download or clone the repository.

git clone https://github.com/kailaisun/GenAI4Urban-Energy.git
cd GenAI4Urban-Energy

Environment Installation

We recommend using Conda (Miniconda) for installation.

Please refer to the UrbanControlnet.

And install segmentation_models.

Dataset Preparation

The dataset is built from publicly available global sources:

Urban boundaries — GHS Urban Centre Database (2023), covering 500 metropolitan areas with 400 m × 400 m grids.
Satellite imagery — Mapbox Static Tiles API.
Population and building data — GHSL P2023A (2020): GHS-BUILT-S, GHS-BUILT-V, GHS-POP.
Environmental constraints — OpenStreetMap, including major roads, water bodies, and railways.

Dataset Download

Download MUSE.

We provide an example for create energy map:

create_building_energy_use_image.py

Dataset preparation

python 5cities-merge_data_preprocess-all.py

GenAI Model Training

We have released our SENSE model checkpoints. You can use it without the steps below.

Otherwise, please refer to the UrbanControlnet.

Finetuning decoder

python E_decodere_train.py
python H_decoder_train.py

Downstream Energy Prediction

For example, we use Segformer as the baseline:

Training on real-world data:

python segformer-image2energy_same_train-real.py

Training on real-world and synthetic data:

python segformer-image2energy_same_train-mix.py

Testing:

python segformer-energy-performance.py

Model evaluation

For computing metric (e.g., FID, SSIM, FSIM, PSNR, etc.), please see our another repo: Evaluation-Metrics

Acknowledgements

This research is supported by the National Research Foundation (NRF), Prime Minister’s Office under its Campus for Research Excellence and Technological Enterprise (CREATE) programme. The Mens, Manus, and Machina (M3S) is an interdisciplinary research group (IRG) of the Massachusetts Institute of Technology and the Singapore MIT Alliance for Research and Technology (SMART) centre.