metadata

tags:
  - clip
library_name: open_clip
pipeline_tag: zero-shot-image-classification
license: mit

Model card for taxabind-vit-b-16

Paper: TaxaBind: A Unified Embedding Space for Ecological Applications

Venue: WACV 2025

Github: https://github.com/mvrl/TaxaBind

TaxaBind

TaxaBind is a multimodal embedding space consisting of six modalities. This model contains image and text modalities in open_clip format. The model is used for zero-shot classification of species images using taxonomic text classes.

Usage

import open_clip

model, preprocess_train, preprocess_val = open_clip.create_model_and_transforms('hf-hub:MVRL/taxabind-vit-b-16')
tokenizer = open_clip.get_tokenizer('hf-hub:MVRL/taxabind-vit-b-16')