ComfyUI-Watermark-Detection

ComfyUI-Watermark-Detection
★ 49

水印检测YOLOv11掩码输出检测可视化
ComfyUI-Watermark-Detection 节点使用 YOLO 模型检测图像水印，输出检测可视化与二值掩码，便于自动屏蔽或标注水印。
💡 在ComfyUI流程中自动检测并生成水印掩码以便后续处理。
🍴 3 Forks💻 Python🔄 2025-05-28
📦
网盘下载
复制链接后前往夸克网盘下载
https://pan.quark.cn/s/e58c8376a81b
📄 README

ComfyUI Watermark Detection Node

This custom node for ComfyUI provides watermark detection capabilities using a YOLO model trained by fancyfeast, the creator of JoyCaption. The model is originally hosted at Hugging Face Space.

Features

  • Detects watermarks in images using YOLOv11
  • Returns both the detection visualization and a binary mask
  • Adjustable confidence threshold for detections

    • Installation

  • Place this folder in your ComfyUI/custom_nodes/ directory
  • The model will be automatically downloaded from Hugging Face on first use

    • Usage

  • Add a Load Watermark Detector node to load the model
  • Connect your image to the Detect Watermarks node
  • Adjust the confidence threshold as needed (default: 0.5)
  • The node outputs:
  • Detection visualization (IMAGE)
  • Binary mask of detected watermarks (MASK)

    • Model Details

  • Model: YOLOv11 trained on watermark detection
  • Original Author: fancyfeast (JoyCaption creator)
  • Model Source: Hugging Face
  • Re-hosted at Hugging Face

    • Notes

  • The model will be downloaded to ComfyUI/models/yolo/ automatically
  • Higher confidence thresholds will result in fewer but more certain detections