# Change-Detection-based Visual Question Answering on QAG-360K (test) Source: [Change-Detection-based Visual Question Answering on QAG-360K (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:38:39.398Z Evaluation of model performance on the test split of the QAG-360K dataset specifically for change-detection-based visual question answering tasks, reporting metrics such as CN, CtW, and OA. 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 | Change-Detection-based Visual Question Answering | | Dataset (including any stated split) | QAG-360K (test) | | Metric | CN | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 14 of 14 | | 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 - [CN](): Higher is better; unit: PERCENT. - [CtW](): Higher is better; unit: PERCENT. - [CfW](): Higher is better; unit: PERCENT. - [IN](): Higher is better; unit: PERCENT. - [DN](): Higher is better; unit: PERCENT. - [LC](): Higher is better; unit: PERCENT. - [SC](): Higher is better; unit: PERCENT. - [CR](): Higher is better; unit: PERCENT. - [AA](): Higher is better; unit: PERCENT. - [OA](): Higher is better; unit: PERCENT. ## Recorded results | Method | CN | Result date | Report paper | | --- | --- | --- | --- | | VisTA | 85.85 | Not recorded | 6955ff9053fdd5c36173e05c | | CDVQA | 82.25 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B-DARFT | 78.13 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B-SFT | 77.74 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B-SFT | 76.13 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B-DARFT | 75.96 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B-DARFT | 75.87 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B-DARFT | 74.65 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B-SFT | 73.53 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B-SFT | 71.65 | Not recorded | 6955ff9053fdd5c36173e05c | | VisTA | 68.55 | Not recorded | 6955ff9053fdd5c36173e05c | | VisTA | 68.29 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-3B | 51.37 | Not recorded | 6955ff9053fdd5c36173e05c | | Qwen2.5-VL-7B | 47.37 | Not recorded | 6955ff9053fdd5c36173e05c | ### VisTA CN: 85.85. Result ID: 6997f99b5698665ec4530ff8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Specialist Methods"}. This record does not assert independent reproduction. ### CDVQA CN: 82.25. Result ID: 6997f99b5698665ec4530ff7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Specialist Methods"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-DARFT CN: 78.13. Result ID: 6997f99b5698665ec4530ffd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Few-shot Fine-tune","Training Samples":"0.5K"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-SFT CN: 77.74. Result ID: 6997f99b5698665ec4530ffc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Few-shot Fine-tune","Training Samples":"0.5K"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-SFT CN: 76.13. Result ID: 6997f99b5698665ec4530fff. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Few-shot Fine-tune","Training Samples":"2K"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-DARFT CN: 75.96. Result ID: 6997f99b5698665ec4531000. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Few-shot Fine-tune","Training Samples":"2K"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-DARFT CN: 75.87. Result ID: 6997f99b5698665ec4531004. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Mean@20 Multi-sample Decoding","Regime":"Few-shot Fine-tune","Training Samples":"2K"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-DARFT CN: 74.65. Result ID: 6997f99b5698665ec4531002. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Mean@20 Multi-sample Decoding","Regime":"Few-shot Fine-tune","Training Samples":"0.5K"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-SFT CN: 73.53. Result ID: 6997f99b5698665ec4531003. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Mean@20 Multi-sample Decoding","Regime":"Few-shot Fine-tune","Training Samples":"2K"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-SFT CN: 71.65. Result ID: 6997f99b5698665ec4531001. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Mean@20 Multi-sample Decoding","Regime":"Few-shot Fine-tune","Training Samples":"0.5K"}. This record does not assert independent reproduction. ### VisTA CN: 68.55. Result ID: 6997f99b5698665ec4530ffb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Few-shot Fine-tune","Training Samples":"0.5K"}. This record does not assert independent reproduction. ### VisTA CN: 68.29. Result ID: 6997f99b5698665ec4530ffe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Few-shot Fine-tune","Training Samples":"2K"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B CN: 51.37. Result ID: 6997f99b5698665ec4530ff9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Zero-shot Baseline"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B CN: 47.37. Result ID: 6997f99b5698665ec4530ffa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Decoding Strategy":"Deterministic Greedy Decoding","Regime":"Zero-shot Baseline"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.