Visual Question Answering On Gqa Test2019

Accuracy

Binary

Consistency

Distribution

Open

Plausibility

Validity

Résultats

Résultats de performance de divers modèles sur ce benchmark

								Paper Title
human	89.3	91.2	98.4	0.0	87.4	97.2	98.9	-
DREAM+Unicoder-VL (MSRA)	76.04	84.46	91.47	3.68	68.6	83.75	96.42	-
TRRNet (Ensemble)	74.03	82.12	89.0	1.29	66.89	83.58	96.76	-
MIL-nbgao	73.81	80.8	91.76	1.7	67.64	83.9	96.73	-
Kakao Brain	73.33	79.68	77.02	2.46	67.73	83.7	96.36	-
Coarse-to-Fine Reasoning, Single Model	72.14	81.16	90.96	2.39	64.19	84.81	96.77	-
270	70.23	77.5	86.94	1.49	63.82	83.77	96.65	-
NSM ensemble (updated)	67.55	80.45	93.83	2.78	56.16	84.16	96.53	-
VinVL-DPT	64.92	82.63	94.37	5.11	49.29	84.91	96.64	-
VinVL+L	64.85	82.59	94.0	4.59	49.19	84.91	96.62	VinVL+L: Enriching Visual Representation with Location Context in VQA
Single Model	64.65	82.63	94.35	4.72	48.77	84.98	96.62	VinVL: Revisiting Visual Representations in Vision-Language Models
Wayne	63.94	80.84	91.54	4.69	49.03	84.74	96.56	-
Single	63.2	77.91	89.84	5.25	50.22	85.15	96.47	-
NSM single (updated)	63.17	78.94	93.25	3.71	49.25	84.28	96.41	-
LXR955, Ensemble	62.71	79.79	93.1	6.42	47.64	85.21	96.36	LXMERT: Learning Cross-Modality Encoder Representations from Transformers
MDETR	62.45	80.91	93.95	5.36	46.15	84.15	96.33	-
1-gqa	62.44	80.28	94.36	5.33	46.69	84.91	96.46	-
UCM	61.49	78.4	88.68	5.7	46.56	84.85	96.33	-
GRN	61.22	78.69	90.31	6.77	45.81	85.43	96.36	Bilinear Graph Networks for Visual Question Answering
lxmert-adv-txt	61.12	78.07	91.13	5.55	46.16	84.8	96.36	-

0 of 127 row(s) selected.