# Visual Question Answering on Scene-VQA easy Source: [Visual Question Answering on Scene-VQA easy]() Collection index: [llms.txt]() Data updated: 2026-03-18T07:23:10.334Z Evaluation of model performance on the easy subset of the Scene-VQA dataset for visual question answering, measured by Accuracy. 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 Question Answering | | Dataset (including any stated split) | Scene-VQA easy | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 2 of 2 | | 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](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | CE-Ensemble | 98 | Not recorded | 68fa78051e209131cea765d8 | | Baseline | 92.5 | Not recorded | 68fa78051e209131cea765d8 | ### CE-Ensemble Accuracy: 98. Result ID: 69ba52c747f34ce924a3cb05. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Distance Metric":"Cosine Distance"}. This record does not assert independent reproduction. ### Baseline Accuracy: 92.5. Result ID: 69ba52c747f34ce924a3cb04. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.