This predictive model can map medium-voltage electrical distribution infrastructure in any country.
0upvotes
29downloads
503views
0
0Shares
updated 14 Nov 2025
Upvote
0
LICENSE MIT
SECTOR Housing, Urban Planning and Infrastructure
About Model
The model functions by applying custom algorithms to publicly available datasets, including NASA satellite imagery, land cover datasets from MODIS, and roadway locations from OpenStreetMap.
Meta has shared its code and a tutorial on how the model is constructed, allowing anyone to replicate it for new countries.
facebookresearch/many-to-many-dijkstra is licensed under the
MIT License
A short and simple permissive license with conditions only requiring preservation of copyright and license notices. Licensed works, modifications, and larger works may be distributed under different terms and without source code.
SegFormer-B5 fine-tuned on 20,000+ Sentinel-2 satellite image chips from 5 districts of Andhra Pradesh, India. Supports land use semantic segmentation (Buildings, Roads, Water Bodies, Open Plots) and bi-temporal change detection. Best mIoU: 0.5474.
Llama 4 Scout 17B-16E-Instruct is a 17B-parameter, instruction-tuned multimodal model developed in collaboration with AI Kosh, Meta and Sarvam. Built on a 16-expert Mixture-of-Experts architecture, it is optimized for high-performance text generation and image understanding, delivering strong efficiency and scalable multimodal reasoning capabilities.
Llama 4 Scout 17B-16E-Instruct is a 17B-parameter, instruction-tuned multimodal model developed in collaboration with AI Kosh, Meta and Sarvam. Built on a 16-expert Mixture-of-Experts architecture, it is optimized for high-performance text generation and image understanding, delivering strong efficiency and scalable multimodal reasoning capabilities.
SegFormer-B5 fine-tuned on 20,000+ Sentinel-2 satellite image chips from 5 districts of Andhra Pradesh, India. Supports land use semantic segmentation (Buildings, Roads, Water Bodies, Open Plots) and bi-temporal change detection. Best mIoU: 0.5474.