CAS-FD: Contact-Aware Temporal Sampling for Single-View Foul vs Dive Recognition

CAS-FD：用于单视角犯规与假摔识别的接触感知时间采样技术

Abstract: Distinguishing a genuine foul from a simulated dive in football remains one of the sport’s most contested fine-grained recognition problems, especially when such decisions have to be from a single broadcast view without multi-view camera angle.

摘要： 在足球比赛中，区分真实的犯规与模拟的假摔仍然是该运动中最具争议的细粒度识别问题之一，尤其是在必须仅凭单一广播视角、缺乏多角度摄像机支持的情况下进行判断时，难度尤为突出。

We introduce a balanced 600-clip single-view Foul/Dive dataset and show that contact-aware sampling concentrating the model’s attention around the moment of physical contact rather than treating all frames equally yields substantially improved recognition of this contact-specific problem.

我们引入了一个包含 600 个片段的平衡单视角“犯规/假摔”数据集，并证明了“接触感知采样”（contact-aware sampling）方法——即让模型将注意力集中在身体接触瞬间，而非同等对待所有帧——能够显著提升针对此类接触相关问题的识别效果。

The proposed approach achieves 86.0% accuracy and macro-F1 0.860 on the held-out test split, a 12 percentage-point gain over contact-unaware alternatives that grows further on unseen data.

该方法在留出的测试集上达到了 86.0% 的准确率和 0.860 的宏观 F1 分数，相比不具备接触感知能力的替代方案提升了 12 个百分点，且在未见数据上的表现优势进一步扩大。

We also evaluate each pipeline component against human annotations, establishing where and why the system succeeds and fails. The result is a documented dataset, a reproducible single-view pipeline, and a grounded evaluation framework for fine-grained contact-event recognition in broadcast football footage.

我们还通过人工标注对流水线的每个组件进行了评估，明确了系统在何处成功、何处失败及其原因。最终成果包括一个有据可查的数据集、一个可复现的单视角处理流水线，以及一个用于足球广播画面中细粒度接触事件识别的扎实评估框架。

The dataset and code are available at this https URL: tamim/contact-aware-dive.

该数据集和代码已在以下链接公开：tamim/contact-aware-dive。