search

dqshuai / MetaFormer

A PyTorch implementation of "MetaFormer: A Unified Meta Framework for Fine-Grained Recognition". A reference PyTorch implementation of “CoAtNet: Marrying Convolution and Attention for All Data Sizes”
MIT License
209 stars 35 forks source link

meta data #22

Open punk1 opened 2 years ago

punk1 commented 2 years ago

Hi, may I ask where did the meta data come from? Time, latitude and longitude are not provided in the dataset

dqshuai commented 2 years ago

Hi, only iNaturalist datasets have time, latitude and longitude, you can get by iNat .

sun-jiao commented 7 months ago

Hi, @dqshuai .

I noticed the MetaFormer-2 model pretrained on iNaturalist 2021 (with meta data) gets a accuracy of 93% on NABirds. And the weights you provided also has meta dimension. I'd like to know how did you evaluate a model with meta dim on a dataset without meta data? Using random meta data or something else?