Files changed (1) hide show
  1. README.md +10 -1
README.md CHANGED
@@ -11,7 +11,16 @@ More details could be found at:
11
  - πŸ† 2026 SoccerNet Challenge - VQA: [DevKit](https://github.com/jyrao/SoccerAgent) β‹… [Rules](https://github.com/jyrao/SoccerAgent/blob/main/Challenge.md) β‹… [Eval (Test)](https://www.codabench.org/competitions/11086/#/results-tab) β‹… [Eval (Challenge)](https://www.codabench.org/competitions/11087/)
12
  - πŸ“‘ Paper Relevant Links: [WebPage](https://jyrao.github.io/SoccerAgent) β‹… [Benchmark](https://huggingface.co/datasets/Homie0609/SoccerBench) β‹… [Database](https://huggingface.co/datasets/Homie0609/SoccerWiki)
13
 
14
- ## Data
 
 
 
 
 
 
 
 
 
15
  Both the test phase and challenge phase are supported by 500 unique QA pairs, which span all 14 aforementioned tasks. You can download the test set and challenge set on this [huggingface page](https://huggingface.co/datasets/SoccerNet/SN-VQA-2026).
16
 
17
  Each QA pair contains three core components in its dictionary:
 
11
  - πŸ† 2026 SoccerNet Challenge - VQA: [DevKit](https://github.com/jyrao/SoccerAgent) β‹… [Rules](https://github.com/jyrao/SoccerAgent/blob/main/Challenge.md) β‹… [Eval (Test)](https://www.codabench.org/competitions/11086/#/results-tab) β‹… [Eval (Challenge)](https://www.codabench.org/competitions/11087/)
12
  - πŸ“‘ Paper Relevant Links: [WebPage](https://jyrao.github.io/SoccerAgent) β‹… [Benchmark](https://huggingface.co/datasets/Homie0609/SoccerBench) β‹… [Database](https://huggingface.co/datasets/Homie0609/SoccerWiki)
13
 
14
+ ##
15
+ You can find the train/valid/test/challenge set of this challenge. With each `.zip` file contains a `.json` file and a `materials` folder, where `.json` file contains the questions (`challenge.zip` has no answers) and `materials` folder contains the required picture/videos in the questions.
16
+
17
+ Such train.zip and valid.zip is split from the benchmark [SoccerBench](https://huggingface.co/datasets/Homie0609/SoccerBench), which is the largest multimodal soccer understanding QA benchmark introduced by our previous work of **SoccerAgent**. In the train/valid dataset, you could find the soccer understanding QA pairs featuring:
18
+
19
+ - **Around 10k QA pairs** across 14 tasks
20
+ - **Balanced distribution** across modalities and difficulty levels
21
+ - **Multi-choice format** with one correct answer and three distractors
22
+ - **Professional annotations** from soccer experts
23
+
24
  Both the test phase and challenge phase are supported by 500 unique QA pairs, which span all 14 aforementioned tasks. You can download the test set and challenge set on this [huggingface page](https://huggingface.co/datasets/SoccerNet/SN-VQA-2026).
25
 
26
  Each QA pair contains three core components in its dictionary: