how to check if an image allready exists for product or manufacturer

[vodiATAD]

I quite sure we must not only insert but also update products
we need an image similarity check before we upload an image

[Sylver11]

@vodiATAD so you would like to have a similarity check on all images of a product and the image locally to check if the image already exists?

[vodiATAD]

lets say, when you update manufacturer.abouts and there are 1 text and 1 image. But at the scraping and extractin with a LLM you find 3 images. You have to update the existing about section with 3 or with 2 images in case one of that images already exists

[vodiATAD]

Idea is to keep a image vector representing the image. in case the same image is added again we can react on that. In minimum we have a checker function like products.image_exists( .. )

[Sylver11]

so I am trying out hash similiarity for the images and it seems to somewhat work. Here is the client:

https://github.com/DAAily/daailyapis-python-client/blob/SPRIN-3417-sparky-create-image-similarity-func/daaily/similarity/client.py

and here is an example with result:

from daaily.lucy import Client
from daaily.lucy.constants import LUCY_V2_BASE_URL_STAGING
from daaily.similarity import Client as SimilarityClient

# Initialize the clients
client = Client(base_url=LUCY_V2_BASE_URL_STAGING)
similarity_client = SimilarityClient()

resp = client.products.get_by_id(20306944)

product = resp.json()

if not product or not product.get("images"):
    raise ValueError("Product has no images.")

similarities = similarity_client.compare_lucy_images(
    product["images"][0]["blob_id"],
    [
        "https://image.architonic.com/pro2-3/20306944/fino---03-smoke-20306944-1-pro-b-arcit18.png",
        "https://image.architonic.com/pfm3-3/2093192/fino-2019-fam-g-arcit18.jpg",
    ],
)

print(similarities)

# {
#     'https://image.architonic.com/pro2-3/20306944/fino---03-smoke-20306944-1-pro-b-arcit18.png': 0, 
#     'https://image.architonic.com/pfm3-3/2093192/fino-2019-fam-g-arcit18.jpg': 38
# }
# The lower the distance between the image the more similar they are
# So a number of 0 means that they are exactly the same

Whereby this is the image that comes from the product and is being passed as the source image: https://storage.googleapis.com/m-on/3100707/product/20306944/nya-nordiska_fino---03-smoke_53780ea9.png

This is the first external image
https://image.architonic.com/pro2-3/20306944/fino---03-smoke-20306944-1-pro-b-arcit18.png

This is the second external image
https://image.architonic.com/pfm3-3/2093192/fino-2019-fam-g-arcit18.jpg

[vodiATAD]

I dont know how to use that feature
"https://image.architonic.com/pro2-3/20306944/fino---03-smoke-20306944-1-pro-b-arcit18.png",
"https://image.architonic.com/pfm3-3/2093192/fino-2019-fam-g-arcit18.jpg",

these URLs may have a logic how to build them. I m , as a API user, dont know it. Why is something needed we dont get out of a product ? Lets take the Blob information from the product instead of URLs we may fail in constructing the URL