Zero-Shot Image Classification
OpenCLIP
English
clip
biology
CV
images
animals
species
taxonomy
rare species
endangered species
evolutionary biology
multimodal
knowledge-guided
Instructions to use imageomics/bioclip with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- OpenCLIP
How to use imageomics/bioclip with OpenCLIP:
import open_clip model, preprocess_train, preprocess_val = open_clip.create_model_and_transforms('hf-hub:imageomics/bioclip') tokenizer = open_clip.get_tokenizer('hf-hub:imageomics/bioclip') - Notebooks
- Google Colab
- Kaggle
Examples
Zero-Shot Classification
pip install torch # whatever version you want
pip install open_clip_torch numpy tqdm torchvision
Suppose you want to evaluate BioCLIP on zero-shot classification on two tasks, <DATASET-NAME> and <DATASET2-NAME>.
You can use examples/zero_shot.py to get top1 and top5 accuracy assuming your tasks are arranged as torchvision's ImageFolder wants.
python examples/zero_shot.py \
--datasets <DATASET-NAME>=<DATASET-FOLDER> <DATASET2-NAME>=<DATASET2-FOLDER>
This will write to logs/bioclip-zero-shot/results.json with your results.