AnyoneMakes

Project profile

FenceFlow

FenceFlow is a fencing competition operations platform, not a fencing game or video. It connects referees, tournament committees, coaches, and spectators.

The submission describes photographing score sheets, using visual OCR to read scores, automatically approving high-confidence results, and sending anomalies to a committee with review reasons. Coaches receive analysis using Bradley–Terry modeling, Monte Carlo simulation, and conversation; spectators receive streamed schedules and commentary.

The repository says the demonstration simulates a workflow using real tournament data. Integrating existing tournament software requires association approval, so the demo is not evidence of live tournament deployment. Recognition, confidence screening, human review, publishing, and coaching tools are described together, but the app, current-versus-submitted version, and claimed results were not independently tested.

AI at runtimeWorkflow automationSoftwareProductivityModeling & simulationVisual recognition

Creator or team (original public names)

Nour Desouki;仓库署名 CozMx

Links & sources

Original postCreator profile

Links may have changed since we checked them. Descriptions from creators have not been independently tested.

Details in the original language

Original title

FenceFlow

Original creator credit

Nour Desouki;仓库署名 CozMx

作品是什么

击剑赛事运营平台,而非击剑游戏或视频作品。

给谁用

赛事委员会、裁判、教练,以及希望查看实时赛果的公众。

问题或体验

纸质手写计分表难辨认、结果录入滞后、错误可能影响排位,裁判协调和观众获取赛况也需要额外人工工作。

如何使用

裁判拍摄计分表,视觉 OCR 识别分数;高置信度表格由 agent 自动批准,异常交委员会复核并记录理由;教练端结合 Bradley-Terry 模型、Monte Carlo 仿真与多轮聊天,公众端流式展示赛程和解说。

得到什么

四类使用界面的赛事运营软件,连接识别、审核、赛果发布和教练分析。仓库说明演示用真实赛事数据模拟,接入现有赛事软件需要协会批准;不能表述为已投入真实赛事运转。

关键机制

OCR 识别、置信度筛选与人工异常复核相连,再把结果送至公众和教练界面;作者描述另有仿真分析。

特点

把裁判、委员会、教练和公众的赛事操作连接起来;现有演示为真实数据模拟,非已投入正式赛事。

Events & award announcements

Built with Opus 4.6: a Claude Code hackathon

Result listed — see original announcement

Original award wording

Finalists

Read original announcement
Project source 1 Project source 2

Sources last checked: 2026-10-11