Image Captioning On Nocaps Entire

评估指标

CIDEr

METEOR

ROUGE-L

SPICE

评测结果

各个模型在此基准测试上的表现结果

									Paper Title	Repository
Lyrics	-	-	-	-	126.8	-	-	-	Lyrics: Boosting Fine-grained Language-Vision Alignment and Comprehension via Semantic-aware Visual Objects	-
GIT, Single Model	88.1	74.81	57.68	37.35	123.39	32.5	63.12	15.94	GIT: A Generative Image-to-text Transformer for Vision and Language
CoCa - Google Brain	87.01	73.71	56.88	37.71	120.55	32.29	62.52	15.47	-	-
Microsoft Cognitive Services team	85.62	71.36	53.62	34.65	114.25	31.27	61.2	14.85	VIVO: Visual Vocabulary Pre-Training for Novel Object Captioning	-
Prismer	84.87	69.99	52.48	33.66	110.84	31.13	60.55	14.91	Prismer: A Vision-Language Model with Multi-Task Experts
Single Model	83.78	68.86	51.06	32.2	110.31	30.55	59.86	14.49	SimVLM: Simple Visual Language Model Pretraining with Weak Supervision
FudanFVL	83.9	68.77	50.84	32.17	108.29	30.64	59.82	14.72	-	-
FudanWYZ	82.95	67.45	49.58	31.38	106.81	30.32	59.18	14.56	-	-
IEDA-LAB	83.25	67.3	48.41	29.27	98.08	28.92	58.56	13.9	-	-
firethehole	80.77	65.55	48.14	30.2	97.61	30.07	58.25	14.74	-	-
vll@mk514	81.61	65.1	46.13	27.32	93.45	28.46	57.4	14.06	-	-
MD	82.43	66.25	47.18	28.2	93.0	28.09	57.57	13.35	-	-
VinVL (Microsoft Cognitive Services + MSR)	81.59	65.15	45.04	26.15	92.46	27.57	56.96	13.07	VinVL: Revisiting Visual Representations in Vision-Language Models
ViTCAP-CIDEr-136.7-ENC-DEC-ViTbfocal10-test-CBS	81.03	64.62	45.26	26.52	87.56	27.36	56.7	12.81	-	-
icgp2ssi1_coco_si_0.02_5_test	79.0	61.95	42.36	24.62	87.34	26.29	55.03	12.01	-	-
evertyhing	78.92	61.6	41.52	23.52	86.0	26.31	54.75	12.1	-	-
Human	76.64	56.46	36.37	19.48	85.34	28.15	52.83	14.67	-	-
RCAL	78.19	60.74	39.11	21.24	82.88	25.72	53.85	12.2	-	-
Oscar	79.57	60.83	38.83	21.02	80.93	25.33	54.07	11.29	-	-
vinvl_yuan_cbs	79.32	60.95	39.5	20.3	79.04	25.44	53.8	11.9	-	-

0 of 39 row(s) selected.