HyperAI초신경

Grounded Situation Recognition On Swig

평가 지표

Top-1 Verb

Top-1 Verb u0026 Value

Top-5 Verbs

Top-5 Verbs u0026 Value

평가 결과

이 벤치마크에서 각 모델의 성능 결과

					Paper Title
Ours (CoFormer+)	58.88	51.10	-	-	Dynamic Scene Understanding from Vision-Language Representations
ClipSitu	58.19	47.23	85.69	68.42	ClipSitu: Effectively Leveraging CLIP for Conditional Predictions in Situation Recognition
CoFormer	44.66	35.98	73.31	57.76	Collaborative Transformers for Grounded Situation Recognition
SituFormer	44.2	35.24	71.21	55.75	Rethinking the Two-Stage Framework for Grounded Situation Recognition
Kernel GraphNet	43.27	35.41	68.72	55.62	Mixture-Kernel Graph Attention Network for Situation Recognition
GSRTR	40.63	32.15	69.81	54.13	Grounded Situation Recognition with Transformers
JSL	39.94	31.44	67.6	51.88	Grounded Situation Recognition
ISL	39.36	30.09	65.51	50.16	Grounded Situation Recognition
CAQ + RE-VGG	38.19	30.23	65.05	50.21	Attention-Based Context Aware Reasoning for Situation Recognition
GraphNet	36.72	27.52	61.90	45.39	Situation Recognition with Graph Neural Networks
RNN + Fusion	35.9	27.45	63.08	46.88	Recurrent Models for Situation Recognition
CRF + Aug	34.12	26.45	62.59	46.88	Commonly Uncommon: Semantic Sparsity in Situation Recognition
CRF	32.34	24.64	58.88	42.76	Situation Recognition: Visual Semantic Role Labeling for Image Understanding

0 of 13 row(s) selected.