# Video Question Answering on Traffic-VQA (test) Source: [Video Question Answering on Traffic-VQA (test)]() Collection index: [llms.txt]() Data updated: 2026-03-12T09:05:03.302Z Evaluation of model performance on the test split of the Traffic-VQA dataset for Video Question Answering. 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) | Traffic-VQA (test) | | Metric | Overall Accuracy (OA) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 38 of 38 | | 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 - [Overall Accuracy (OA)](): Higher is better; unit: PERCENT. - [Traffic Cognition Accuracy](): Higher is better; unit: PERCENT. - [Traffic Perception Accuracy](): Higher is better; unit: PERCENT. - [Vehicle Violation Accuracy](): Higher is better; unit: PERCENT. - [Vehicle Behavior Accuracy](): Higher is better; unit: PERCENT. - [Vehicle Accuracy](): Higher is better; unit: PERCENT. - [Pedestrian Violation Accuracy](): Higher is better; unit: PERCENT. - [Pedestrian Behavior Accuracy](): Higher is better; unit: PERCENT. - [Pedestrian Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Overall Accuracy (OA) | Result date | Report paper | | --- | --- | --- | --- | | Qwen3-VL-8B + CTCNet | 83.156 | Not recorded | 69b248b3d030084c22e2fada | | Qwen3-VL-8B | 79.598 | Not recorded | 69b248b3d030084c22e2fada | | GPT-4o | 67.72 | Not recorded | 69b248b3d030084c22e2fada | | Gemini-2.5-Flash | 67.47 | Not recorded | 69b248b3d030084c22e2fada | | GPT-4o | 64.81 | Not recorded | 69b248b3d030084c22e2fada | | GPT-4o | 63.66 | Not recorded | 69b248b3d030084c22e2fada | | CTCNet | 61.94 | Not recorded | 69b248b3d030084c22e2fada | | Gemini-2.5-Pro | 61.41 | Not recorded | 69b248b3d030084c22e2fada | | Gemini-2.5-Flash | 59.51 | Not recorded | 69b248b3d030084c22e2fada | | Gemini-2.5-Flash | 57.01 | Not recorded | 69b248b3d030084c22e2fada | | Gemini-2.5-Pro | 56.17 | Not recorded | 69b248b3d030084c22e2fada | | Gemini-2.5-Pro | 51.5 | Not recorded | 69b248b3d030084c22e2fada | | Qwen3-VL-8B | 47.62 | Not recorded | 69b248b3d030084c22e2fada | | Qwen3-VL-8B | 47.56 | Not recorded | 69b248b3d030084c22e2fada | | CTCNet | 47.56 | Not recorded | 69b248b3d030084c22e2fada | | MiniCPM-V | 45.37 | Not recorded | 69b248b3d030084c22e2fada | | MiniCPM-V | 45.27 | Not recorded | 69b248b3d030084c22e2fada | | GeoChat | 44.22 | Not recorded | 69b248b3d030084c22e2fada | | Qwen3-VL-8B | 44.09 | Not recorded | 69b248b3d030084c22e2fada | | CTCNet | 44.09 | Not recorded | 69b248b3d030084c22e2fada | | GeoChat | 44 | Not recorded | 69b248b3d030084c22e2fada | | Qwen2.5-VL-7B | 43.73 | Not recorded | 69b248b3d030084c22e2fada | | Qwen2.5-VL-7B | 43.69 | Not recorded | 69b248b3d030084c22e2fada | | MiniCPM-V | 43.62 | Not recorded | 69b248b3d030084c22e2fada | | GeoPix | 43.62 | Not recorded | 69b248b3d030084c22e2fada | | GeoPix | 43.59 | Not recorded | 69b248b3d030084c22e2fada | | GeoChat | 43.08 | Not recorded | 69b248b3d030084c22e2fada | | GeoPix | 42.73 | Not recorded | 69b248b3d030084c22e2fada | | DeepSeek-VL | 42.64 | Not recorded | 69b248b3d030084c22e2fada | | DeepSeek-VL | 42.5 | Not recorded | 69b248b3d030084c22e2fada | | Qwen2.5-VL-7B | 40.57 | Not recorded | 69b248b3d030084c22e2fada | | DeepSeek-VL | 39.2 | Not recorded | 69b248b3d030084c22e2fada | | MiniGPT-v2 | 36.68 | Not recorded | 69b248b3d030084c22e2fada | | MiniGPT-v2 | 36.04 | Not recorded | 69b248b3d030084c22e2fada | | Falcon | 34.8 | Not recorded | 69b248b3d030084c22e2fada | | MiniGPT-v2 | 34.47 | Not recorded | 69b248b3d030084c22e2fada | | Falcon | 34.22 | Not recorded | 69b248b3d030084c22e2fada | | Falcon | 33.59 | Not recorded | 69b248b3d030084c22e2fada | ### Qwen3-VL-8B + CTCNet Overall Accuracy (OA): 83.156. Result ID: 69b281a36dbb7075b95d51df. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"MUL"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Overall Accuracy (OA): 79.598. Result ID: 69b281a36dbb7075b95d51de. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"MUL","Variant":"Finetuned","Venue":"Ali 2025"}. This record does not assert independent reproduction. ### GPT-4o Overall Accuracy (OA): 67.72. Result ID: 69b281a36dbb7075b95d51da. [Reporting paper]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"OpenAI 2024"}. This record does not assert independent reproduction. ### Gemini-2.5-Flash Overall Accuracy (OA): 67.47. Result ID: 69b281a36dbb7075b95d51db. [Reporting paper]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"Google 2025"}. This record does not assert independent reproduction. ### GPT-4o Overall Accuracy (OA): 64.81. Result ID: 69b281a36dbb7075b95d51c2. [Reporting paper]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"OpenAI 2024"}. This record does not assert independent reproduction. ### GPT-4o Overall Accuracy (OA): 63.66. Result ID: 69b281a36dbb7075b95d51ce. [Reporting paper]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"OpenAI 2024"}. This record does not assert independent reproduction. ### CTCNet Overall Accuracy (OA): 61.94. Result ID: 69b281a36dbb7075b95d51dd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"MUL","Variant":"Few-Shot"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Overall Accuracy (OA): 61.41. Result ID: 69b281a36dbb7075b95d51dc. [Reporting paper]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"Google 2025"}. This record does not assert independent reproduction. ### Gemini-2.5-Flash Overall Accuracy (OA): 59.51. Result ID: 69b281a36dbb7075b95d51c3. [Reporting paper]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"Google 2025"}. This record does not assert independent reproduction. ### Gemini-2.5-Flash Overall Accuracy (OA): 57.01. Result ID: 69b281a36dbb7075b95d51cf. [Reporting paper]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"Google 2025"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Overall Accuracy (OA): 56.17. Result ID: 69b281a36dbb7075b95d51c4. [Reporting paper]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"Google 2025"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Overall Accuracy (OA): 51.5. Result ID: 69b281a36dbb7075b95d51d0. [Reporting paper]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"Google 2025"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Overall Accuracy (OA): 47.62. Result ID: 69b281a36dbb7075b95d51d9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"MUL","Variant":"Base","Venue":"Ali 2025"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Overall Accuracy (OA): 47.56. Result ID: 69b281a36dbb7075b95d51c1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"OPT","Variant":"Base","Venue":"Ali 2025"}. This record does not assert independent reproduction. ### CTCNet Overall Accuracy (OA): 47.56. Result ID: 69b281a36dbb7075b95d51c5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"OPT","Variant":"Few-Shot"}. This record does not assert independent reproduction. ### MiniCPM-V Overall Accuracy (OA): 45.37. Result ID: 69b281a36dbb7075b95d51d2. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"arXiV 2024"}. This record does not assert independent reproduction. ### MiniCPM-V Overall Accuracy (OA): 45.27. Result ID: 69b281a36dbb7075b95d51ba. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"arXiV 2024"}. This record does not assert independent reproduction. ### GeoChat Overall Accuracy (OA): 44.22. Result ID: 69b281a36dbb7075b95d51d6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"CVPR 2024"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Overall Accuracy (OA): 44.09. Result ID: 69b281a36dbb7075b95d51cd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"TIR","Variant":"Base","Venue":"Ali 2025"}. This record does not assert independent reproduction. ### CTCNet Overall Accuracy (OA): 44.09. Result ID: 69b281a36dbb7075b95d51d1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"TIR","Variant":"Few-Shot"}. This record does not assert independent reproduction. ### GeoChat Overall Accuracy (OA): 44. Result ID: 69b281a36dbb7075b95d51be. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"CVPR 2024"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B Overall Accuracy (OA): 43.73. Result ID: 69b281a36dbb7075b95d51d8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"Ali 2025"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B Overall Accuracy (OA): 43.69. Result ID: 69b281a36dbb7075b95d51c0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"Ali 2025"}. This record does not assert independent reproduction. ### MiniCPM-V Overall Accuracy (OA): 43.62. Result ID: 69b281a36dbb7075b95d51c6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"arXiV 2024"}. This record does not assert independent reproduction. ### GeoPix Overall Accuracy (OA): 43.62. Result ID: 69b281a36dbb7075b95d51d5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"GRSM 2025"}. This record does not assert independent reproduction. ### GeoPix Overall Accuracy (OA): 43.59. Result ID: 69b281a36dbb7075b95d51bd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"GRSM 2025"}. This record does not assert independent reproduction. ### GeoChat Overall Accuracy (OA): 43.08. Result ID: 69b281a36dbb7075b95d51ca. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"CVPR 2024"}. This record does not assert independent reproduction. ### GeoPix Overall Accuracy (OA): 42.73. Result ID: 69b281a36dbb7075b95d51c9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"GRSM 2025"}. This record does not assert independent reproduction. ### DeepSeek-VL Overall Accuracy (OA): 42.64. Result ID: 69b281a36dbb7075b95d51bc. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"arXiV 2024"}. This record does not assert independent reproduction. ### DeepSeek-VL Overall Accuracy (OA): 42.5. Result ID: 69b281a36dbb7075b95d51d4. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"arXiV 2024"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B Overall Accuracy (OA): 40.57. Result ID: 69b281a36dbb7075b95d51cc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"Ali 2025"}. This record does not assert independent reproduction. ### DeepSeek-VL Overall Accuracy (OA): 39.2. Result ID: 69b281a36dbb7075b95d51c8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"arXiV 2024"}. This record does not assert independent reproduction. ### MiniGPT-v2 Overall Accuracy (OA): 36.68. Result ID: 69b281a36dbb7075b95d51d3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"arXiv 2023"}. This record does not assert independent reproduction. ### MiniGPT-v2 Overall Accuracy (OA): 36.04. Result ID: 69b281a36dbb7075b95d51bb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"arXiv 2023"}. This record does not assert independent reproduction. ### Falcon Overall Accuracy (OA): 34.8. Result ID: 69b281a36dbb7075b95d51d7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"MUL","Venue":"arXiV 2025"}. This record does not assert independent reproduction. ### MiniGPT-v2 Overall Accuracy (OA): 34.47. Result ID: 69b281a36dbb7075b95d51c7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"arXiv 2023"}. This record does not assert independent reproduction. ### Falcon Overall Accuracy (OA): 34.22. Result ID: 69b281a36dbb7075b95d51bf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"OPT","Venue":"arXiV 2025"}. This record does not assert independent reproduction. ### Falcon Overall Accuracy (OA): 33.59. Result ID: 69b281a36dbb7075b95d51cb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality":"TIR","Venue":"arXiV 2025"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.