AI 人脸检测
通过 OpenCV DNN 使用 YuNet ONNX。可单独切换每张人脸，保留选定人物的清晰画面。
为不可逆脱敏而设计
一切都在您的机器上运行。导出永久改写像素 — 编辑器无法揭掉模糊层还原原图。
手动跟踪与静态区域
绘制框并使用 CSRT 跟踪，或在指定时间段锁定矩形区域。
手动音频区间
在时间轴上标记静音或蜂鸣。无语音识别 — 音频仅由操作员划定范围。
马赛克、模糊或纯色
选择符合政策的样式。每种方式都会在导出中改写像素 — 不是图像工具可剥离的叠加层。
离线优先设计
本网站和应用工作流均无上传管道。适用于学校、法律和安全场景。
Capabilities
Each capability has its own page so searchers and procurement can land on the exact workflow they named.
Face redaction
Auto-detect faces and draw manual boxes. Keep selected people clear; burn the rest.
Blur license plate
Auto-detect plates and draw manual boxes. Track and burn mosaics for CCTV, bodycam, and dashcam.
PII video redaction
A checklist of identifiers compliance teams actually mask in footage.
Audio redaction
Mark mute or beep ranges by hand. No speech recognition — audio is operator-timed only.
Offline & on-premise
No upload path. Process on an air-gapped Windows PC after activation.
AI video redaction
Automated detection so you are not painting every frame by hand.
Video anonymization
Research and ethics-review language for irreversible local anonymisation.
Video redaction software
Procurement brief: deployment, masking styles, and how we differ from cloud SaaS.