ModelListedExternal

Watermark Remover

6 specialized UNet++ segmentation models for watermark detection, combined with pixel-wise max fusion.

Sourcehuggingface.co/christophernavas/watermark-remover/tree/509fb6833823f4e9c519f8e4f87a9053567c779c

KindOtherVersionRevision 509fb68

Model source

Kind: Other
File size: 80.1 MB
Version: vRevision 509fb68
Updated: Sep 27, 2026
Downloads: 0
Tasks: Image segmentation
Source: Hugging Face

Source description

Watermark Remover — Multi-Model v2

6 specialized UNet++ segmentation models for watermark detection, combined with pixel-wise max fusion.

Models

File	Specialty	Encoder	Best Val IoU	Size
segmenter_centered_text.pth	centered_text	efficientnet-b4	N/A	84.0 MB
segmenter_line_pattern.pth	line_pattern	efficientnet-b4	N/A	84.0 MB
segmenter_logo.pth	logo	efficientnet-b4	N/A	84.0 MB
segmenter_overlay_text.pth	overlay_text	efficientnet-b4	N/A	84.0 MB
segmenter_repeated_text.pth	repeated_text	efficientnet-b4	N/A	84.0 MB
segmenter_tiny_corner.pth	tiny_corner	efficientnet-b4	N/A	84.0 MB

Architecture

Image → [Model A, B, C, D, E, F] → pixel-wise max(masks) → LaMa inpainting → Clean image

Each model is trained on synthetic watermarks matching its specialty. Fusion strategy: pixel_max (logical OR of all masks).

Usage

import segmentation_models_pytorch as smp
import torch

model = smp.UnetPlusPlus(encoder_name="efficientnet-b4", encoder_weights=None, in_channels=3, classes=1)
state_dict = torch.load("segmenter_repeated_text.pth", map_location="cpu")
model.load_state_dict(state_dict)
model.eval()

License

Apache-2.0

Author

DevynLabs — AI tools for creators.

Sources

1 source

Hugging FacePrimaryVersion 509fb6833823f4e9c519f8e4f87a9053567c779c

huggingface.co/christophernavas/watermark-remover/tree/509fb6833823f4e9c519f8e4f87a9053567c779c

Verified Sep 27

Model artifacts

9 artifacts

segmenter_centered_text.pth

pth · 80.1 MB · SHA-256 bcaadb84ef76…5a49 · Hugging Face

Download

segmenter_gemini.pth

pth · 80.1 MB · SHA-256 e89c02a6baa3…8205 · Hugging Face

Download

segmenter_line_pattern.pth

pth · 80.1 MB · SHA-256 0268cccdec88…7edf · Hugging Face

Download

segmenter_logo.pth

pth · 80.1 MB · SHA-256 c6a399fdd1a1…6da0 · Hugging Face

Download

segmenter_overlay_text.pth

pth · 80.1 MB · SHA-256 717c938b9553…f380 · Hugging Face

Download

segmenter_repeated_text.pth

pth · 80.1 MB · SHA-256 bfd26cffb00d…8d0e · Hugging Face

Download

segmenter_tiny_corner.pth

pth · 80.1 MB · SHA-256 618212981978…fa83 · Hugging Face

Download

segmenter_universal.pth

pth · 80.1 MB · SHA-256 3ffcda89cfd5…3426 · Hugging Face

Download

segmenter.pth

pth · 80.1 MB · SHA-256 a4579c690e43…5a66 · Hugging Face

Download

Source excerpts

2 excerpts

ReadmeSep 27huggingface.co/christophernavas/watermark-remover

--- license: apache-2.0 tags: - watermark-removal - image-segmentation - unet++ - pytorch --- # Watermark Remover — Multi-Model v2 6 specialized UNet++ segmentation models for watermark detection, combined with pixel-wise max fusion. ## Models | File | Specialty | Encoder | Best Val IoU | Size | |------|-----------|---------|-------------|------| | segmenter_centered_text.pth | centered_text | efficientnet-b4 | N/A | 84.0 MB | | segmenter_line_pattern.pth | line_pattern | efficientnet-b4 | N/A | 84.0 MB | | segmenter_logo.pth | logo | efficientnet-b4 | N/A | 84.0 MB | | segmenter_overlay_text.pth | overlay_text | efficientnet-b4 | N/A | 84.0 MB | | segmenter_repeated_text.pth | repeated_text | efficientnet-b4 | N/A | 84.0 MB | | segmenter_tiny_corner.pth | tiny_corner | efficientnet-b4 | N/A | 84.0 MB | ## Architecture ``` Image → [Model A, B, C, D, E, F] → pixel-wise max(masks) → LaMa inpainting → Clean image ``` Each model is trained on synthetic watermarks matching its specialty. Fusion strategy: `pixel_max` (logical OR of all masks). ## Usage ```python import segmentation_models_pytorch as smp import torch model = smp.UnetPlusPlus(encoder_name="efficientnet-b4", encoder_weights=None, in_channels=3, classes=1) state_dict = torch.load("segmenter_repeated_text.pth", map_location="cpu") model.load_state_dict(state_dict) model.eval() ``` ## License Apache-2.0 ## Author [DevynLabs](https://devynlabs.com) — AI tools for creators.

NoteSep 27huggingface.co/christophernavas/watermark-remover

Source context: 0 downloads · 7 likes · Pipeline image-segmentation · Repo christophernavas/watermark-remover