/ZeroShotEval

A Python toolkit for evaluating the quality of classification models for tasks of Zero-Shot Learning (ZSL) including the models construction with different insides (e.g. miscellaneous NN architectures), procedure for training and saving models for further evaluation.

Primary LanguagePythonApache License 2.0Apache-2.0

ZeroShotEval

A Python toolkit for evaluating the quality of classification models for tasks of Zero-Shot Learning (ZSL) including the models construction with different insides (e.g. miscellaneous NN architectures), procedure for training and saving models for further evaluation.

Pipeline

zeroshoteval_pipeline

NOTE: in the figure above there are examples of Python scripts and its names. Please note that it is work names for visualizing the structure of the project.

Usage

Step 1: datasets loading

Go to scripts directory and run downloading scripts. Example:

cd ./scripts 
./cub_download.sh

All datasets are placed in data directory