Search papers, labs, and topics across Lattice.
University of British, Columbia University of Copenhagen, Carleton University University of Copenhagen Vector Institute Vector Institute, Taylor Geospatial University of Copenhagen University of British Columbia Vector Institute
5
0
7
Earth observation streams and foundation model embeddings can be compressed 1,800脳 into continuous neural fields while retaining 90%+ downstream performance and cutting data retrieval latency tenfold.
Foundation models can dramatically enhance the classification of critical infrastructure, achieving up to 57.9% accuracy compared to traditional methods.
GeoFMs are overconfident and equally fragile under stress, regardless of whether they are pretrained on EO or ImageNet data, challenging conventional evaluation practices.
Ditch mean pooling in your geospatial foundation models: richer pooling methods like GeM can boost accuracy by up to 5% and slash the geographic generalization gap by 40%.
