EmbodiedScan

Visit Tool

EmbodiedScan is an open-source AI Agents & Automation tool that provides a holistic multi-modal 3D perception suite for embodied AI. It includes datasets, benchmarks, and a framework for 3D scene understanding.

Claim this tool

1View

At a glance

Pricing

Open Source

Free tier

Yes

API

Skill level

Technical

About

What is EmbodiedScan?

EmbodiedScan is a comprehensive open-source suite designed for holistic multi-modal 3D perception, specifically tailored for embodied AI. It addresses the gap in traditional research by focusing on ego-centric 3D scene understanding from first-person observations. The suite includes EmbodiedScan, a multi-modal, ego-centric 3D perception dataset with over 5k scans, 1M RGB-D views, 1M language prompts, and 160k 3D-oriented boxes across 760 categories. It also features the Embodied Perceptron baseline framework, capable of processing arbitrary multi-modal inputs for robust 3D perception and language-grounded tasks. The project provides tools for data processing, benchmark evaluation, and model adaptations, making it a valuable resource for researchers and developers in computer vision and robotics.

Best used for

Ideal for developers and data scientists who need to build and evaluate embodied AI agents capable of understanding complex 3D environments from a first-person perspective. Especially valuable for researchers working on multi-modal perception, language grounding, and robotic navigation in simulated or real-world scenarios.

Common actions

develop embodied AI

understand 3D scenes

process multi-modal data

benchmark perception models

research robotics

open-sourceautomated workflowcollaborationdeepfakeworkflowslow-code/no-codegithub copilotface swapping"AI Agents"

Capabilities

Key features

Multi-modal 3D perception
Ego-centric 3D dataset
Language-grounded tasks
Embodied Perceptron framework
Benchmark evaluation tools
Pretrained models

Target Audience

developerdata scientiststartup founder

Integrations

Not yet documented

Pricing & Plans

Open Source

Free

FAQs

What kind of data is included in the EmbodiedScan dataset?

The EmbodiedScan dataset includes over 5,000 scans, 1 million ego-centric RGB-D views, 1 million language prompts, and 160,000 3D-oriented bounding boxes across 760 categories. It also provides dense semantic occupancy with 80 common categories, designed for holistic 3D scene understanding.

What is the Embodied Perceptron framework?

The Embodied Perceptron is a baseline framework introduced with EmbodiedScan. It is designed to process an arbitrary number of multi-modal inputs, demonstrating strong 3D perception capabilities for fundamental 3D perception tasks and language-grounded tasks, both within benchmarks and in real-world applications.

How can I get access to the EmbodiedScan data and models?

EmbodiedScan is open-source and available on GitHub. You can clone the repository to access the code. For downloading the full datasets, including EmbodiedScan v2 beta and MMScan, you typically need to fill out a form to apply for access, as mentioned in the project's news section.

Trending

Subcategories trending in AI Agents & Automation

AI Frameworks & Infra Chatbots & Conversational AI Workflow Agents Personal Assistants RAG & Document AI Voice Agents

Trending

Also listed in

This tool also appears in

Content & Design › 3D & Animation

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