scieee Science in your language
[en] (orig)

AI-assisted image and video annotation with BIIGLE

Author: Zurowietz, Martin; Stoye, Jens; Nattkemper, Tim W.
Publisher: Zenodo
DOI: 10.5281/zenodo.17708536
Source: https://zenodo.org/records/17708536/files/biigle-poster-joint-plenary-2025.pdf
AI-assis ed image and ideo
anno a ion wi h BIIGLE
BIIGLE
Wha is BIIGLE?
Image and ideo da a is essen ial o
biodi e si y esea ch and he ea h
sciences. This da a is in e p e ed and
analysed ia image and ideo
anno a ion, i.e. egions o in e es and
objec s a e ma ked and desc ibed.
Anno a ing la ge image and ideo
da ase s is e y ime-consuming. Thus,
specialized so wa e is necessa y o
make his p ocess as e�cien as
possible. BIIGLE is such a so wa e and
p o ides ools o manual and AI-
assis ed anno a ion, collabo a ion,
eaching and mo e.
sea s a
A polygon image anno a ion
BIIGLE is open sou ce so wa e and openly
de eloped a Gi Hub. A public BIIGLE
ins ance is ope a ed a Biele eld Uni e si y,
suppo ed by he de.NBI cloud and pa o
he NFDI4Biodi e si y se ice ca alog. The
ins ance is a ailable a biigle.de and used
by housands o use s wo ldwide.
biigle.de
BIIGLE Communi y
Lea n mo e
38.000.000
5.500.0004.700
40
AI-assis ance in BIIGLE
in [email protected]
gi hub.com/biigle
The Machine lea ning
Assis ed Image Anno a ion
(MAIA) ool helps o build a
aining da ase o machine
lea ning and hen pe o ms
au oma ic objec de ec ion
on new da ase s. I p o ides
an e�cien use in e ace o
e iew he de ec ions and
selec labels o desc ibe he
de ec ed objec s.
The LabelBOT ool sugges s
labels o new anno a ions in
eal ime. Sugges ions a e
based on exis ing anno a ions
in he en i e BIIGLE da abase.
This g ea ly educes he
amoun o use in e ac ions
equi ed o manual
anno a ion.
The Magic SAM ool
gene a es a de ailed polygon
anno a ion o an objec
wi h a single click.
Combined wi h LabelBOT,
manual use in e ac ions a e
educed o a minimum.
The La go ool o�e s
powe ul so ing and
fil e ing mechanisms.
Anno a ions can be
so ed by "simila i y"
o by "ou lie s" o
quickly iden i y and fix
anno a ion e o s.
De ec ion
Classifica ion
Segmen a ion
Re iew