# Fine-grained Multi-Image Understanding on VLM2-Bench Source: [Fine-grained Multi-Image Understanding on VLM2-Bench]() Collection index: [llms.txt]() Data updated: 2026-04-27T07:23:32.489Z Evaluation of fine-grained multi-image understanding capabilities using the VLM2-Bench dataset, measured across detailed sub-metrics like Matching (Mat), Counting (Cnt), and Grouping (Grp) for objects and persons. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Fine-grained Multi-Image Understanding | | Dataset (including any stated split) | VLM2-Bench | | Metric | Matching Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 18 of 18 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Matching Score](): Higher is better; unit: PERCENT. - [Tracking Score](): Higher is better; unit: PERCENT. - [Counting Score (Obj)](): Higher is better; unit: PERCENT. - [Correspondence Score (Obj)](): Higher is better; unit: PERCENT. - [Grouping Score (Obj)](): Higher is better; unit: PERCENT. - [Counting Score (Per)](): Higher is better; unit: PERCENT. - [Correspondence Score (Per)](): Higher is better; unit: PERCENT. - [Grouping Score (Per)](): Higher is better; unit: PERCENT. - [Average Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Matching Score | Result date | Report paper | | --- | --- | --- | --- | | Human-Level | 95.06 | Not recorded | 69eeed6355a3c32af89f95fa | | MiCo-7B | 57.14 | Not recorded | 69eeed6355a3c32af89f95fa | | MM-Eureka-7B | 55.6 | Not recorded | 69eeed6355a3c32af89f95fa | | CGC-8B | 48.85 | Not recorded | 69eeed6355a3c32af89f95fa | | VLAA-Thinker-7B | 47.49 | Not recorded | 69eeed6355a3c32af89f95fa | | CGC-7B | 42.11 | Not recorded | 69eeed6355a3c32af89f95fa | | NoisyRollout-7B | 40.93 | Not recorded | 69eeed6355a3c32af89f95fa | | ThinkLite-VL-7B | 40.45 | Not recorded | 69eeed6355a3c32af89f95fa | | Qwen2.5-VL-7B | 35.91 | Not recorded | 69eeed6355a3c32af89f95fa | | InternVL2.5-26B | 30.5 | Not recorded | 69eeed6355a3c32af89f95fa | | Qwen3-VL-8B | 27.03 | Not recorded | 69eeed6355a3c32af89f95fa | | InternVL3-8B | 23.94 | Not recorded | 69eeed6355a3c32af89f95fa | | InternVL2.5-8B | 21.24 | Not recorded | 69eeed6355a3c32af89f95fa | | LLaVA-Video-7B | 18.53 | Not recorded | 69eeed6355a3c32af89f95fa | | Qwen2-VL-7B | 18.07 | Not recorded | 69eeed6355a3c32af89f95fa | | mPLUG-Owl2-7B | 17.37 | Not recorded | 69eeed6355a3c32af89f95fa | | LLaVA-OV-7B | 16.6 | Not recorded | 69eeed6355a3c32af89f95fa | | LongVQA-7B | 14.29 | Not recorded | 69eeed6355a3c32af89f95fa | ### Human-Level Matching Score: 95.06. Result ID: 69ef0edd47f34ce924d4ded4. [Reporting paper]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### MiCo-7B Matching Score: 57.14. Result ID: 69ef0edd47f34ce924d4dee3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"Qwen2.5-VL"}. This record does not assert independent reproduction. ### MM-Eureka-7B Matching Score: 55.6. Result ID: 69ef0edd47f34ce924d4dedf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"Qwen2.5-VL"}. This record does not assert independent reproduction. ### CGC-8B Matching Score: 48.85. Result ID: 69ef0edd47f34ce924d4dee5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"Qwen3-VL"}. This record does not assert independent reproduction. ### VLAA-Thinker-7B Matching Score: 47.49. Result ID: 69ef0edd47f34ce924d4dee2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"Qwen2.5-VL"}. This record does not assert independent reproduction. ### CGC-7B Matching Score: 42.11. Result ID: 69ef0edd47f34ce924d4dee4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"Qwen2.5-VL"}. This record does not assert independent reproduction. ### NoisyRollout-7B Matching Score: 40.93. Result ID: 69ef0edd47f34ce924d4dee0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"Qwen2.5-VL"}. This record does not assert independent reproduction. ### ThinkLite-VL-7B Matching Score: 40.45. Result ID: 69ef0edd47f34ce924d4dee1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"Qwen2.5-VL"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B Matching Score: 35.91. Result ID: 69ef0edd47f34ce924d4dedd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### InternVL2.5-26B Matching Score: 30.5. Result ID: 69ef0edd47f34ce924d4deda. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Matching Score: 27.03. Result ID: 69ef0edd47f34ce924d4dede. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### InternVL3-8B Matching Score: 23.94. Result ID: 69ef0edd47f34ce924d4dedb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### InternVL2.5-8B Matching Score: 21.24. Result ID: 69ef0edd47f34ce924d4ded9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### LLaVA-Video-7B Matching Score: 18.53. Result ID: 69ef0edd47f34ce924d4ded6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### Qwen2-VL-7B Matching Score: 18.07. Result ID: 69ef0edd47f34ce924d4dedc. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### mPLUG-Owl2-7B Matching Score: 17.37. Result ID: 69ef0edd47f34ce924d4ded8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### LLaVA-OV-7B Matching Score: 16.6. Result ID: 69ef0edd47f34ce924d4ded5. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. ### LongVQA-7B Matching Score: 14.29. Result ID: 69ef0edd47f34ce924d4ded7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base MLLM":"-"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.