BEiTImageEncoder
BEiTImageEncoder is an image encoder that wraps the image embedding functionality using the BEiT models by Microsoft. The models have been implemented in HuggingFace transformers here
Supported Models:
- microsoft/beit-base-patch16-224-pt22k-ft22k
- microsoft/beit-base-patch16-224
- microsoft/beit-base-patch16-224-pt22k
- microsoft/beit-base-patch16-384
- microsoft/beit-large-patch16-224-pt22k-ft22k
- microsoft/beit-large-patch16-512
- microsoft/beit-large-patch16-224-pt22k
- microsoft/beit-large-patch16-224
- microsoft/beit-large-patch16-384
Usage
via Docker image (recommended)
from jina import Flow
f = Flow().add(uses='jinahub+docker://BEiTImageEncoder')via source code
from jina import Flow
f = Flow().add(uses='jinahub://BEiTImageEncoder')