1 paper
Kaizhen Tan, Yuantao Deng
Pretrained vision embeddings are increasingly used as general-purpose representations for modelling how people appraise urban scenes, and are validated almost entirely by how well…