Open-vocabulary Attribute Detection

Showing image 267 of 2000 (COCO id 50165) with 1 / 0 objects and 14 / 91 / 12 attributes.

Objects can be from the base or the novel set. Attributes can be positive, negative or unknown.

Objects

Attributes

Positive Negative Unknown

Object 1: train

cleanliness	clean / neat
color quantity	multicolored / colorful
color	green
color	white
group	single / one
length	short
material	metal / metallic
optical property	opaque
pattern	plain
position	horizontal / lying
size	small / little
state	on
texture	smooth / sleek
tone	dark
cleanliness	unclean / dirt
clothes color	black
clothes color	blue
clothes color	brown
clothes color	gray
clothes color	green
clothes color	orange
clothes color	pink
clothes color	red
clothes color	tan
clothes color	violet
clothes color	white
clothes color	yellow
clothes pattern	dotted / speckled
clothes pattern	floral
clothes pattern	lettered
clothes pattern	plaid / tartan
clothes pattern	plain
clothes pattern	striped / lined
clothes pattern	tiled
color quantity	single-colored / unicolored
color quantity	two-colored
cooked	cooked / baked
cooked	raw / fresh
face expression	angry / mad
face expression	disgust / frowning
face expression	fear / surprise
face expression	happy / smiling
face expression	neutral / calm
face expression	sad / unhappy
face expression	sleepy / sleeping
gender	female / woman
gender	male / man
group	group / bunch
hair color	black
hair color	blue
hair color	brown
hair color	gray
hair color	green
hair color	orange
hair color	pink
hair color	red
hair color	tan
hair color	violet
hair color	white
hair color	yellow
hair length	bald
hair length	long
hair length	short
hair tone	dark
hair tone	light / bright
hair type	curly / curled
hair type	straight
length	long
material	asphalt / cement
material	ceramic / brick
material	glass
material	leather
material	paper / cardboard
material	polymers / plastic
material	stone / granite
material	textile / cloth
material	wood / wooden
maturity	adult / old
maturity	young / baby
optical property	reflective
optical property	transparent / translucent
pattern	dotted / speckled
pattern	floral
pattern	lettered
pattern	plaid / tartan
pattern	striped / lined
pattern	tiled
position	sitting / sit
position	vertical / upright
size	big / large
state	closed
state	covered
state	cracked
state	dilapidated / ruined
state	dry
state	empty
state	folded / bend
state	full / whole
state	off
state	open
state	piece / cut
state	wet
texture	rough
texture	soft / fluffy
tone	light / bright
color	black
color	blue
color	brown
color	gray
color	orange
color	pink
color	red
color	tan
color	violet
color	yellow
order	messy / disordered
order	ordered / arranged

Open-vocabulary Attribute Detection

Dataset Visualization

Objects

Attributes

Object 1: train