Netzgemeinde Hubzilla
Anmelden
Registrieren
Netzgemeinde Hubzilla
Anmelden
Registrieren
Channel Apps
Kanal
Über
Fotos
Dateien
Kalender
Artikel
Wikis
System Apps
Fehler melden
Hilfe
QRator
Sprache
Suche
Verzeichnis
Zufälliger Kanal
Wed, 26 Jun 2024 07:35:30 +0200
Profil ansehen
Jupiter Rowland
jupiter_rowland@hub.netzgemeinde.eu
Found another image-describing AI to outdo with my own image descriptions; CW: long (over 800 characters), image description meta, not praising AI
Artikel ansehen
Zusammenfassung ansehen
I've got two free image-describing AIs at hand now to play with and pit against my own research and writing.
One, a demo of the
Large Language and Vision Assistant
(LLaVA).
Two, the
ASU EdPlus Image Accessibility Generator
based on GPT-4o.
The former has produced train wrecks twice already. I haven't tried the latter yet, but I don't think it'll stand a chance.
Sorry to say to all you AI fans who are fully convinced that AI can describe any possible image out there with the same confidence, accuracy and informativity as even human experts in whichever field the image covers. But I've got too many unfair advantages over AI.
#
Long
#
LongPost
#
CWLong
#
CWLongPost
#
AltText
#
AltTextMeta
#
CWAltTextMeta
#
ImageDescription
#
ImageDescriptions
#
ImageDescriptionMeta
#
CWImageDescriptionMeta
#
LLaVA
#
GPT4o
#
AI
Fediverse
2
Dirk Blank :nonazis: 🇩🇪
Chief335
Link zur Quelle
Konversationsmerkmale
Lädt...
Konversationsmerkmale
Lädt...
