Task-Informed Meta-Learning applied to crop classification, using the cropharvest dataset.
This repository mirrors the structure of the cropharvest benchmarks - specifically, we train a meta-model in the TIML Learner, and use the state dictionary saved by the learner in a finetuning loop.
Anaconda running python 3.6 is used as the package manager. To get set up with an environment, install Anaconda from the link above, and (from this directory) run
conda env create -f environment.yml
This will create an environment named timl-classification
with all the necessary packages to run the code. To
activate this environment, run
conda activate timl-classification
The main script to train the models is then deep_learning.py
, with the model configurations controlled by the config
The trained TIML model is available on Zenodo.