This image is directly related to the image photo-search. It provides an efficient tool to index photos to later be queried by the services offered in the photo-search image.
This image therefore replaces the old inefficient indexing script created with Python.
Running the indexing tool in the image requires the --model parameter. The model path should be set to the 0_CLIPModel directory of the clip-ViT-B-32 Model. A read-only copy of the model is sufficient, and in fact the only file required to be present is:
./model.safetensorsAll other files can be ignored / skipped / deleted. This means that the overall size of the model and related files is about 570 - 580 MiB.
The tool needs read-only access to the photos in order to index them. This is done through the --photos parameter.
The web server from the photo-search image has an internal indexing server. The indexing tool sends its data to this server for persisting with the vector database. You need to specify the base address of the indexing server with the --indexing-server parameter. If not specified, the default value of http://localhost:8081 is used.
You can control the size of each batch of images indexed by the tool. Use the --batch-size parameter for this. By default, a batch size of 20 is used.
Add the --help argument when running the container to get more help.
More information can be found in the git repo
Content type
Image
Digest
sha256:49c128a9f…
Size
3.6 MB
Last updated
about 1 month ago
docker pull rokeller/photo-search-indexing:v0.6.10