Referring Expression Comprehension on KVG-Bench benchmark leaderboard · SOTA2 Research

Referring Expression Comprehension on KVG-Bench

81.33Accuracy (Air, Seen Categories)Human + search

Evaluation Results

Method	Links
Human + search2025.03		81.33	85.56	68	74.17	86.67	78.03	78.89	74	74.17	84.44	86.67	79.63	78.83
Qwen3-VL-32B2025.03		72.37	68.27	63.45	48.21	77.14	63.38	87.2	48.89	51.79	66.29	81.93	66.18	64.52
KARL2025.03		71.05	83.65	70.34	45.02	87.86	67.3	85.6	50.37	69.64	80.9	67.47	70.04	68.41
Qwen2-VL-72B2025.03		63.16	75.96	59.31	40.24	77.14	59.34	80.8	42.96	59.82	65.17	66.27	62.32	60.55
InternVL2-76B2025.03		62.5	74.04	60	41.04	76.43	59.22	78.4	51.11	56.25	43.82	55.42	57.9	58.68
Qwen3-VL-8B2025.03		61.84	68.27	55.17	40.24	75	56.94	73.6	44.44	42.86	65.17	66.27	57.54	57.19
Human2025.03		59.33	66.67	50.84	44.17	65.33	57.27	48.33	45.33	51.67	64.45	68	55.56	56.41
UniVG-R12025.03		57.24	74.04	52.41	35.06	78.57	55.3	73.6	40	55.36	60.67	67.46	58.46	56.59
DeepSeek-VL22025.03		51.32	60.57	53.1	29.08	63.57	47.98	62.4	35.56	50.89	44.94	39.76	47.06	47.6
Qwen2-VL-7B2025.03		48.03	74.04	51.3	33.07	65.71	50.38	76	33.33	54.46	57.3	59.04	55.33	52.4
CogVLM-G2025.03		46.71	64.42	49.66	34.26	63.57	48.61	79.2	31.11	54.46	56.18	66.27	56.43	51.8
DINO-X2025.03		43.42	49.04	42.76	28.29	41.43	38.89	62.4	35.56	48.21	31.46	49.4	45.77	41.69
YOLO-World2025.03		41.45	28.85	8.28	14.74	30.71	23.36	50.4	2.22	24.11	1.12	3.61	17.83	21.11
G-DINO-1.6-Pro2025.03		39.47	41.35	48.97	23.11	24.29	33.59	44	40	39.29	32.58	27.71	37.68	35.25
Visual-RFT2025.03		36.18	51.92	32.41	20.7	47.86	34.72	52.8	21.48	31.25	46.07	44.58	38.24	36.15
Perception-R12025.03		23.03	32.69	14.48	12.35	37.14	21.84	41.6	11.85	26.79	16.85	28.92	25.18	23.2
Shikra-7B2025.03		20.39	25.96	15.17	16.33	28.57	20.33	51.2	19.26	25	16.85	22.89	27.94	23.43