ACM SIGSPATIAL 2025
Last week, Lin Che had the pleasure of attending ACM SIGSPATIAL 2025, held in Minneapolis.
He presented the paper “๐๐ง๐ฌ๐ฎ๐ฉ๐๐ซ๐ฏ๐ข๐ฌ๐๐ ๐๐ซ๐๐๐ง ๐๐๐ง๐ ๐๐ฌ๐ ๐๐๐ฉ๐ฉ๐ข๐ง๐ ๐ฐ๐ข๐ญ๐ก ๐๐ญ๐ซ๐๐๐ญ ๐๐ข๐๐ฐ ๐๐จ๐ง๐ญ๐ซ๐๐ฌ๐ญ๐ข๐ฏ๐ ๐๐ฅ๐ฎ๐ฌ๐ญ๐๐ซ๐ข๐ง๐ ๐๐ง๐ ๐ ๐๐๐จ๐ ๐ซ๐๐ฉ๐ก๐ข๐๐๐ฅ ๐๐ซ๐ข๐จ๐ซ”
In this work, we proposed the first unsupervised approach for urban land use classification and mapping based on street view images. Contrastive Clustering with Geographical Priors (CCGP) framework enhances clustering performance and ensures greater spatial coherence of the resulting maps.
๐ Paper: https://arxiv.org/abs/2504.17551
๐ป Code: https://github.com/lin102/CCGP
