# Sports Video Understanding on DeepSport (test) Source: [Sports Video Understanding on DeepSport (test)]() Collection index: [llms.txt]() Data updated: 2026-03-13T06:58:00.126Z Evaluation of fine-grained sports video understanding capabilities, covering recognition, rule assessment, coaching feedback, and live commentary generation on the DeepSport test set. 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 | Sports Video Understanding | | Dataset (including any stated split) | DeepSport (test) | | Metric | Fine-Grained Recognition Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 13 of 13 | | 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 - [Fine-Grained Recognition Accuracy](): Higher is better; unit: PERCENT. - [Rule & Procedural Accuracy](): Higher is better; unit: PERCENT. - [Assessment & Coaching Accuracy](): Higher is better; unit: PERCENT. - [Live Commentary & Reporting Accuracy](): Higher is better; unit: PERCENT. - [Overall Average Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Fine-Grained Recognition Accuracy | Result date | Report paper | | --- | --- | --- | --- | | DeepSport | 49.89 | Not recorded | 691bfe116750be83fe5cec65 | | Qwen3-VL-8B-Thinking | 46.75 | Not recorded | 691bfe116750be83fe5cec65 | | GPT-5 | 46.5 | Not recorded | 691bfe116750be83fe5cec65 | | InternVL3.5-14B | 46.38 | Not recorded | 691bfe116750be83fe5cec65 | | Baseline + Sports Curriculum SFT | 46.11 | Not recorded | 691bfe116750be83fe5cec65 | | Qwen3-VL-235B-A22B-Thinking | 44.96 | Not recorded | 691bfe116750be83fe5cec65 | | Qwen3-VL-8B-Instruct | 43.82 | Not recorded | 691bfe116750be83fe5cec65 | | Baseline + SFT | 40.67 | Not recorded | 691bfe116750be83fe5cec65 | | Video-R1 | 38.21 | Not recorded | 691bfe116750be83fe5cec65 | | T\* | 35.79 | Not recorded | 691bfe116750be83fe5cec65 | | VideoITG | 35.39 | Not recorded | 691bfe116750be83fe5cec65 | | SoccerChat | 31.43 | Not recorded | 691bfe116750be83fe5cec65 | | Qwen2.5-VL-7B-Instruct (Baseline) | 28.2 | Not recorded | 691bfe116750be83fe5cec65 | ### DeepSport Fine-Grained Recognition Accuracy: 49.89. Result ID: 69b3b4aa47f34ce9249d1f5d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"9.81"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Thinking Fine-Grained Recognition Accuracy: 46.75. Result ID: 69b3b4aa47f34ce9249d1f53. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### GPT-5 Fine-Grained Recognition Accuracy: 46.5. Result ID: 69b3b4aa47f34ce9249d1f51. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### InternVL3.5-14B Fine-Grained Recognition Accuracy: 46.38. Result ID: 69b3b4aa47f34ce9249d1f55. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### Baseline + Sports Curriculum SFT Fine-Grained Recognition Accuracy: 46.11. Result ID: 69b3b4aa47f34ce9249d1f5c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"8.72"}. This record does not assert independent reproduction. ### Qwen3-VL-235B-A22B-Thinking Fine-Grained Recognition Accuracy: 44.96. Result ID: 69b3b4aa47f34ce9249d1f56. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Instruct Fine-Grained Recognition Accuracy: 43.82. Result ID: 69b3b4aa47f34ce9249d1f52. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### Baseline + SFT Fine-Grained Recognition Accuracy: 40.67. Result ID: 69b3b4aa47f34ce9249d1f5b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"10"}. This record does not assert independent reproduction. ### Video-R1 Fine-Grained Recognition Accuracy: 38.21. Result ID: 69b3b4aa47f34ce9249d1f54. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### T\* Fine-Grained Recognition Accuracy: 35.79. Result ID: 69b3b4aa47f34ce9249d1f57. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### VideoITG Fine-Grained Recognition Accuracy: 35.39. Result ID: 69b3b4aa47f34ce9249d1f58. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### SoccerChat Fine-Grained Recognition Accuracy: 31.43. Result ID: 69b3b4aa47f34ce9249d1f59. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B-Instruct (Baseline) Fine-Grained Recognition Accuracy: 28.2. Result ID: 69b3b4aa47f34ce9249d1f5a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.