# 3D Visual Question Answering on 3DMV-VQA (held-out) Source: [3D Visual Question Answering on 3DMV-VQA (held-out)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:39:44.180Z Evaluation of model performance on the held-out split of the 3DMV-VQA dataset for 3D Visual Question Answering, utilizing metrics covering Concept, Counting, Relation, Comparison, and Overall performance. 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 | 3D Visual Question Answering | | Dataset (including any stated split) | 3DMV-VQA (held-out) | | Metric | Concept Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 10 of 10 | | 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 - [Concept Accuracy](): Higher is better; unit: PERCENT. - [Counting Accuracy](): Higher is better; unit: PERCENT. - [Relation Accuracy](): Higher is better; unit: PERCENT. - [Comparison Accuracy](): Higher is better; unit: PERCENT. - [Overall Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Concept Accuracy | Result date | Report paper | | --- | --- | --- | --- | | 3D-LLM | 68.9 | Not recorded | 68fa70261e209131cea49662 | | 3D-LLM | 68.1 | Not recorded | 68fa70261e209131cea49662 | | 3D-CLR | 66.1 | Not recorded | 68fa70261e209131cea49662 | | 3D-LLM | 63.4 | Not recorded | 68fa70261e209131cea49662 | | BLIP | 61.9 | Not recorded | 68fa70261e209131cea49662 | | 3D-Feature+LSTM | 61.2 | Not recorded | 68fa70261e209131cea49662 | | flamingo | 60 | Not recorded | 68fa70261e209131cea49662 | | NS-VQA | 59.8 | Not recorded | 68fa70261e209131cea49662 | | flamingo | 58.7 | Not recorded | 68fa70261e209131cea49662 | | BLIP | 58 | Not recorded | 68fa70261e209131cea49662 | ### 3D-LLM Concept Accuracy: 68.9. Result ID: 69910c733a6c314849054b11. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"flamingo"}. This record does not assert independent reproduction. ### 3D-LLM Concept Accuracy: 68.1. Result ID: 69910c733a6c314849054b13. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"BLIP2-flanT5"}. This record does not assert independent reproduction. ### 3D-CLR Concept Accuracy: 66.1. Result ID: 69910c733a6c314849054b0c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"explicit object representations and neuro-symbolic reasoning":true}. This record does not assert independent reproduction. ### 3D-LLM Concept Accuracy: 63.4. Result ID: 69910c733a6c314849054b12. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"BLIP2-opt"}. This record does not assert independent reproduction. ### BLIP Concept Accuracy: 61.9. Result ID: 69910c733a6c314849054b10. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Input":"MultiView"}. This record does not assert independent reproduction. ### 3D-Feature+LSTM Concept Accuracy: 61.2. Result ID: 69910c733a6c314849054b0b. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### flamingo Concept Accuracy: 60. Result ID: 69910c733a6c314849054b0e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Input":"MultiView"}. This record does not assert independent reproduction. ### NS-VQA Concept Accuracy: 59.8. Result ID: 69910c733a6c314849054b0a. [Reporting paper]() Recorded evaluation configuration: {"explicit object representations and neuro-symbolic reasoning":true}. This record does not assert independent reproduction. ### flamingo Concept Accuracy: 58.7. Result ID: 69910c733a6c314849054b0d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Input":"SingleImage"}. This record does not assert independent reproduction. ### BLIP Concept Accuracy: 58. Result ID: 69910c733a6c314849054b0f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Input":"SingleImage"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.