My project scope is to being able to confirm if the image is visually identical with another image with minimal false positives / false negatives so it can be used in for confirming the image pairs in automatic processing. . For example, to confirm that some image in Wikimedia Commons is same than some image in GLAM archives. One usecase is preventing duplicates when uploading, but also knowing when updating metadata of the image that bot is updating the correct image.

In T368500#10127917, @Prototyperspective wrote:

Only shows "Search ready" after uploading a small jpeg file via Browse.

Confirmed that if preprosessing is toggled off and scaling is changed to LANCZOS the resulted image hashes are identical to Python imagehash library's phash.

libfastimagehash is a imagehash replacement for C/C++ ... The methods used to compute the image hashes are identical to the imagehash python library, however, due to some slight differences in the way OpenCV vs Pillow images are resized, the final image hashes are not always exactly the same.

• https://github.com/simon987/fastimagehash

Some stats

After comparing the Imagehash.py and OpenCV imagehash

• https://github.com/JohannesBuchner/imagehash/blob/master/imagehash/__init__.py
• https://github.com/opencv/opencv_contrib/blob/4.x/modules/img_hash/src/phash.cpp#L13

Some comments and thoughs

@JJMC89 Thank you for your answer, problem solved!

I updated the SQL query so that it will fetch large numbers instead of zeroes. (ie. SELECT max(phash) FROM imagehash) Also screenshot for consistency.

I got an error when i tried to query bigint columns

I am still interested for archive comments as it makes possible to for example analyse if there were notability discussion before page was deleted and how many users participated to discussion with edit. (ie. starting discussion finnish wikipedia it generates semi-standard comment lines which can be filtered using SQL)

I created a ticket T367945 for superset being down. Should we modify the test or wait if it will come back up?

@Xqt Yes, I looked those as an example when I started.

It would be useful to have a block that would block unregistered users and new users but would not affect users who have at least 100 edits in any Wikimedia wikis and are not blocked in any Wikimedia wikis. This would allow the blocking of IP ranges but still allow regular users to edit.

Ok, now i tested the adding user to the project. So shell name is case sensitive and it worked when i filled only one of the values.

Hi Andrew,

Is there any timetable when this is fixed (or is there alternative methods to add project members)?

No, in Finnish Wikipedia it is that there is data is "no available" except the IPv4/IPv6 information. In some rare cases there is country information.

Missing information is confirmed in Finnish Wikipedia also.

Do you know @taavi @bd808 Who/how i should ask to add 'https://imagehash-sparql.wmcloud.org/sparql' to query.wikidata.org queryserver whitelist? (preferably in Hackathon)

url string field for thumbnails is now replaced with thumb_width (int) and thumb_type (enum) fields and url is dropped from db.

@Nardog: Can you be more specific what kind of quirks are with <button> tag?

Piilotetut ja mallineiden kautta tulevat luokat voi ohittaa.