Scene Graph Prediction with Limited Labels

ICCV 2019 Vincent S. ChenParoma VarmaRanjay KrishnaMichael BernsteinChristopher ReLi Fei-Fei

Visual knowledge bases such as Visual Genome power numerous applications in computer vision, including visual question answering and captioning, but suffer from sparse, incomplete relationships. All scene graph models to date are limited to training on a small set of visual relationships that have thousands of training labels each... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT LEADERBOARD
Predicate Classification ImageCLEF-DA LimLabel (Categ. + Spat.) [email protected] 47.53 # 1
[email protected] 45.49 # 1
[email protected] 47.04 # 1
Scene Graph Detection ImageCLEF-DA LimLabel (Categ. + Spat.) [email protected] 19.28 # 1
[email protected] 18.69 # 1
Scene Graph Classification ImageCLEF-DA LimLabel (Categ. + Spat.) [email protected] 21.44 # 1
[email protected] 20.91 # 1
[email protected] 21.34 # 1
Scene Graph Detection Visual Genome LimLabel (Categ. + Spat.) [email protected] 8.64 # 1
[email protected] 4.04 # 1
[email protected] 6.75 # 1
Scene Graph Classification Visual Genome LimLabel (Categ. + Spat.) [email protected] 14.16 # 1
[email protected] 12.69 # 1
[email protected] 13.91 # 1
Predicate Classification Visual Genome LimLabel (Categ. + Spat.) [email protected] 28.53 # 1
[email protected] 24.72 # 1
[email protected] 27.76 # 1
Scene Graph Detection VRD LimLabel (Categ. + Spat.) [email protected] 17.67 # 1