InfoBayAI/Narrated_Eocentric_Dataset

Narrated Egocentric Video Dataset

Overview

The Narrated Egocentric Video Dataset is an egocentric video collection intended for applications involving user-centric visual interactions, action understanding, spatial reasoning, and object-level task understanding.

The dataset contains video content in .mov and .mp4 formats. Its subject areas are relevant to research and development involving computer vision, video understanding, embodied AI, robotics, and intelligent agent training.

The dataset catalogue describes egocentric interactions that can support work on spatial reasoning, action understanding, object manipulation, and task understanding.

Further details about the narration format, transcript availability, annotation structure, and metadata schema are to be confirmed.

Dataset Highlights

• Egocentric Video Content: First-person-perspective video data.
• Video Formats: .mov and .mp4.
• Spatial Reasoning: Relevant to research involving user-centric spatial interactions.
• Action Understanding: Relevant to understanding actions and activities in video.
• Object-Level Interactions: Covers themes involving object interactions, manipulation, and task understanding.
• Agent Training: Relevant to research and development involving intelligent agents.

Dataset Features

Feature	Description
Dataset type	Egocentric video
Video formats	MOV, MP4
Perspective	Egocentric / first-person

Subject and Task Coverage

The dataset catalogue identifies the following areas of interest:

1. User-centric interactions for spatial reasoning.
2. Action understanding.
3. Agent training.
4. Object-level interactions.
5. Manipulation and task understanding.

Supported AI Tasks

Potential research and development areas suggested by the dataset's stated themes include:

• Egocentric video understanding
• Action understanding
• Spatial reasoning
• Object interaction analysis
• Object manipulation research
• Task understanding
• Embodied AI research
• Intelligent agent development

Applications

The dataset's subject areas may be relevant to:

• First-person video analysis
• Computer vision research
• Video understanding systems
• Robotics and embodied AI research
• Spatial reasoning systems
• Object interaction and manipulation research
• Intelligent agent development
• AI and machine learning experimentation

Industries and Research Areas

The dataset catalogue presents a broad range of industry sectors across its egocentric video collection, including:

• Agriculture, landscaping, and miscellaneous activities
• Arts, crafts, and handmade production
• Automotive, repair, and retail operations
• Construction and facilities management
• Consumer electronics and wearables manufacturing
• Domestic services and home support
• Electronic components and PCB assembly
• Food preparation and kitchen support
• Food service and culinary activities
• General manufacturing and material handling
• Metalworking and plastics
• Manufacturing facilities and factory operations
• Optical products manufacturing and inspection
• Packaging, logistics, and warehousing
• Quality inspection and finishing
• Retail and customer service
• Textile, apparel, and accessories manufacturing
• Toy manufacturing

Dataset Format

The confirmed video file extensions are:

• .mov
• .mp4

Dataset Structure

The exact directory and Parquet schema have not yet been verified.

The following is a conceptual illustration, not a confirmed representation of the actual files:

Narrated_Egocentric_Dataset/
├── video_samples/
    ├── sample_001.mp4
    └── sample_002.mov

Intended Audience

This dataset may be relevant to:

• AI and machine learning researchers
• Computer vision teams
• Video analytics developers
• Robotics and embodied AI teams
• Autonomous systems researchers
• Academic institutions
• AI product development teams
• Enterprise AI and research organizations

Dataset Statistics

Dataset-specific statistics have not been confirmed.

The broader egocentric video catalogue screenshot displays the following collection-level figures:

Catalogue metric	Displayed value
Countries covered	11
Total duration	228,189 hours
Total tasks	132

Notes and Limitations

• The video formats .mov and .mp4 have been confirmed.
• The exact narration type and storage format have not been established.
• Transcript and annotation availability has not been confirmed.
• The Parquet file's columns and relationship to the video files have not yet been inspected.
• Catalogue-level statistics and industry distributions should not be interpreted as dataset-specific figures without verification.
• No license or access terms are asserted in this README.

Summary

The Narrated Egocentric Video Dataset is an egocentric video collection in MOV and MP4 formats, with catalogue themes covering spatial reasoning, action understanding, object-level interactions, manipulation, task understanding, and agent training.

It is intended to be relevant to research and development in computer vision, video understanding, robotics, embodied AI, and intelligent agent systems.

Data Creation

Procured through formal agreements and generated in the ordinary course of business.

Considerations

This dataset is provided for research and educational purposes only. It contains only sample data. For access to the full dataset and enterprise licensing options, please visit our website InfoBay.AI or contact us directly.

-Ph: +91 9236398619
-Email: datareq@infobay.ai