DIS

Visit Tool

DIS is an open-source computer vision tool for highly accurate dichotomous image segmentation. It provides models and datasets for research and general use in image editing, 3D modeling, and art design.

Claim this tool

No Views Yet

At a glance

Pricing

Open Source

Free tier

Yes

API

Skill level

Technical

About

What is DIS?

DIS (Dichotomous Image Segmentation) is an open-source project providing a framework for highly accurate image segmentation. Developed by Xuebin Qin et al. and presented at ECCV 2022, it focuses on a newly formulated DIS task. The repository includes the IS-Net architecture, sample datasets (DIS5K V1.0), and pre-trained models for both academic comparisons and general use. While DIS V1.0 has limitations, a more comprehensive DIS V2.0 dataset and model are under development. The tool is designed for researchers and developers interested in advanced image segmentation, with applications spanning 3D modeling, image editing, and art design.

Best used for

Ideal for researchers and developers who need to implement advanced image segmentation, evaluate model performance with specialized metrics, and integrate segmentation capabilities into applications. Especially valuable for academic projects and open-source contributions in computer vision.

Common actions

perform image segmentation

develop computer vision models

process image datasets

workflowslow-code/no-codeautomated workflowopen-sourcecollaborationdeepfakeface swappinggithub copilot"AI Agents"

Capabilities

Key features

Dichotomous image segmentation
IS-Net architecture
DIS5K dataset V1.0
Pre-trained models
Human Correction Efforts metric
Gradio web demo

Target Audience

researchersdevelopersacademicsai/ml engineers

Integrations

hugging-face

Pricing & Plans

Open Source

Free

FAQs

What is the difference between DIS V1.0 and DIS V2.0 datasets?

DIS V1.0 (DIS5K) includes samples from over 200 categories but lacks robustness for common real-world objects like humans or animals. DIS V2.0 is a comprehensive, large-scale dataset under development, designed to cover more categories with extremely well-annotated samples for improved model performance.

How can I try DIS without setting up the environment locally?

You can try DIS through a free demo provided on Bohrium Notebook. Additionally, there is a Web Demo integrated into Huggingface Spaces using Gradio, which allows for easy interaction and testing of the DIS model directly in your browser.

Are there pre-trained models available for general use?

Yes, an optimized model named 'isnet-general-use.pth' is available for general use, which is distinct from the academic version trained on DIS V1.0. This model can be downloaded and used with the provided simple inference.py code on your own datasets.

Trending

Subcategories trending in Content & Design

AI Writing Assistants Audio & Music Video Generation Photo Editing Graphic Design Video Editing

Trending

Explore

Browse AI tools by category

Content & Design Productivity & Business Coding & Development AI Agents & Automation Research & Education Wellness & Lifestyle Career Development Marketing & Growth Data & Analytics Customer Support & CX Finance E-commerce