# Video Question Answering on AGQA Balanced indirect references 2.0 (test) Source: [Video Question Answering on AGQA Balanced indirect references 2.0 (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:58:48.077Z Evaluation of model performance on the test split of the AGQA Balanced indirect references 2.0 dataset for Video Question Answering. Metrics focus on detailed Precision and Recall across Object, Action, and Temporal reasoning components. 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 | Video Question Answering | | Dataset (including any stated split) | AGQA Balanced indirect references 2.0 (test) | | Metric | Precision (Object-B) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 4 of 4 | | 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 - [Precision (Object-B)](): Higher is better; unit: PERCENT. - [Precision (Object-O)](): Higher is better; unit: PERCENT. - [Precision (Object-All)](): Higher is better; unit: PERCENT. - [Precision (Action-B)](): Higher is better; unit: PERCENT. - [Precision (Action-O)](): Higher is better; unit: PERCENT. - [Precision (Action-All)](): Higher is better; unit: PERCENT. - [Precision (Temporal-B)](): Higher is better; unit: PERCENT. - [Precision (Temporal-O)](): Higher is better; unit: PERCENT. - [Precision (Temporal-All)](): Higher is better; unit: PERCENT. - [Recall (Object-B)](): Higher is better; unit: PERCENT. - [Recall (Object-O)](): Higher is better; unit: PERCENT. - [Recall (Object-All)](): Higher is better; unit: PERCENT. - [Recall (Action-B)](): Higher is better; unit: PERCENT. - [Recall (Action-O)](): Higher is better; unit: PERCENT. - [Recall (Action-All)](): Higher is better; unit: PERCENT. - [Recall (Temporal-B)](): Higher is better; unit: PERCENT. - [Recall (Temporal-O)](): Higher is better; unit: PERCENT. - [Recall (Temporal-All)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Precision (Object-B) | Result date | Report paper | | --- | --- | --- | --- | | SHG-VQA | 76.93 | Not recorded | 68fa6f4b1e209131cea44d66 | | PSAC | 63.69 | Not recorded | 68fa6f4b1e209131cea44d66 | | HME | 62.95 | Not recorded | 68fa6f4b1e209131cea44d66 | | HCRN | 54.06 | Not recorded | 68fa6f4b1e209131cea44d66 | ### SHG-VQA Precision (Object-B): 76.93. Result ID: 6990b52a3a6c314849029fa9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### PSAC Precision (Object-B): 63.69. Result ID: 6990b52a3a6c314849029fa6. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### HME Precision (Object-B): 62.95. Result ID: 6990b52a3a6c314849029fa7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### HCRN Precision (Object-B): 54.06. Result ID: 6990b52a3a6c314849029fa8. [Reporting paper]() [Method origin paper]() [Implementation]() 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.