Spaces:
Running
Set up BioCLIP 2 app code based on BioCLIP demo
Using example images from the first BioCLIP demo, updating the code to run inference with BioCLIP 2 and uses pybioclip and select TreeOfLife-200M sample images.
Due to space persistent storage limitations, embeddings are fetched from the TreeOfLife-200M repo and metadata for the images comes from demo-data. The images will be retrieved from an S3 bucket, as with the original bioclip demo.
@vimar
and
@thompsonmj
, this should function (without returning images) once we have the embeddings merged into TreeOfLife-200M (this PR) and bioclip-2/metadata.parquet
is added to the demo-data repo (using download to local folder to address persistent storage restrictions).
Image return will have to wait for the S3 bucket to be setup, but the code is in place.
Parquet file turned out smaller than expected, so I added it too. Images are currently being uploaded, but otherwise this should be ready to go.