# Visual Grounding on HR-Bench 4K Source: [Visual Grounding on HR-Bench 4K]() Collection index: [llms.txt]() Data updated: 2026-06-16T09:34:43.173Z Evaluation of Visual Grounding performance on the HR-Bench 4K dataset, reporting specific accuracy metrics for Single, Cross, and Average conditions. 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 | Visual Grounding | | Dataset (including any stated split) | HR-Bench 4K | | Metric | Accuracy (Single) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 17 of 17 | | 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 - [Accuracy (Single)](): Higher is better; unit: PERCENT. - [Accuracy (Cross)](): Higher is better; unit: PERCENT. - [Accuracy (Average)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (Single) | Result date | Report paper | | --- | --- | --- | --- | | LazyMCoT | 96.2 | Not recorded | 6a30d8ffd33dfb23e783d12e | | HiDe | 95.5 | Not recorded | 6a30d8ffd33dfb23e783d12e | | LazyMCoT | 94.3 | Not recorded | 6a30d8ffd33dfb23e783d12e | | DeepEyes | 91.3 | Not recorded | 6a30d8ffd33dfb23e783d12e | | Thyme-VL | 91 | Not recorded | 6a30d8ffd33dfb23e783d12e | | Base (Qwen3-VL-8B-Instruct) | 90.5 | Not recorded | 6a30d8ffd33dfb23e783d12e | | ViCrop | 90.5 | Not recorded | 6a30d8ffd33dfb23e783d12e | | DeepScan | 90.1 | Not recorded | 6a30d8ffd33dfb23e783d12e | | LazyMCoT | 89.5 | Not recorded | 6a30d8ffd33dfb23e783d12e | | TreeVGR | 89.5 | Not recorded | 6a30d8ffd33dfb23e783d12e | | Dyfo | 89.2 | Not recorded | 6a30d8ffd33dfb23e783d12e | | Base (Qwen2.5-VL-7B-Instruct) | 88.8 | Not recorded | 6a30d8ffd33dfb23e783d12e | | HiDe | 88.5 | Not recorded | 6a30d8ffd33dfb23e783d12e | | ZoomRefine | 88.5 | Not recorded | 6a30d8ffd33dfb23e783d12e | | ViCrop | 88 | Not recorded | 6a30d8ffd33dfb23e783d12e | | Base (InternVL3-8B-Instruct) | 82.8 | Not recorded | 6a30d8ffd33dfb23e783d12e | | GPT-4o | 70 | Not recorded | 6a30d8ffd33dfb23e783d12e | ### LazyMCoT Accuracy (Single): 96.2. Result ID: 6a31189dfbc373592c2dccd2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### HiDe Accuracy (Single): 95.5. Result ID: 6a31189dfbc373592c2dccd1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### LazyMCoT Accuracy (Single): 94.3. Result ID: 6a31189dfbc373592c2dccc8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### DeepEyes Accuracy (Single): 91.3. Result ID: 6a31189dfbc373592c2dccca. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"false"}. This record does not assert independent reproduction. ### Thyme-VL Accuracy (Single): 91. Result ID: 6a31189dfbc373592c2dcccb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"false"}. This record does not assert independent reproduction. ### Base (Qwen3-VL-8B-Instruct) Accuracy (Single): 90.5. Result ID: 6a31189dfbc373592c2dccc7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B-Instruct","Training Free":"false"}. This record does not assert independent reproduction. ### ViCrop Accuracy (Single): 90.5. Result ID: 6a31189dfbc373592c2dcccd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### DeepScan Accuracy (Single): 90.1. Result ID: 6a31189dfbc373592c2dccd0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### LazyMCoT Accuracy (Single): 89.5. Result ID: 6a31189dfbc373592c2dccc6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InternVL3-8B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### TreeVGR Accuracy (Single): 89.5. Result ID: 6a31189dfbc373592c2dcccc. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"false"}. This record does not assert independent reproduction. ### Dyfo Accuracy (Single): 89.2. Result ID: 6a31189dfbc373592c2dccce. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### Base (Qwen2.5-VL-7B-Instruct) Accuracy (Single): 88.8. Result ID: 6a31189dfbc373592c2dccc9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"false"}. This record does not assert independent reproduction. ### HiDe Accuracy (Single): 88.5. Result ID: 6a31189dfbc373592c2dccc5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InternVL3-8B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### ZoomRefine Accuracy (Single): 88.5. Result ID: 6a31189dfbc373592c2dcccf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### ViCrop Accuracy (Single): 88. Result ID: 6a31189dfbc373592c2dccc4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InternVL3-8B-Instruct","Training Free":"true"}. This record does not assert independent reproduction. ### Base (InternVL3-8B-Instruct) Accuracy (Single): 82.8. Result ID: 6a31189dfbc373592c2dccc3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InternVL3-8B-Instruct","Training Free":"false"}. This record does not assert independent reproduction. ### GPT-4o Accuracy (Single): 70. Result ID: 6a31189dfbc373592c2dccc2. [Reporting paper]() Recorded evaluation configuration: {"Training Free":"true"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.