ThisIsPepper/EgocentricSpecialties

Egocentric Specialties Dataset

A dataset of egocentric (first-person) and top-down video footage of skilled jewelry manufacturing workers.

If you would are a robotics company looking to source this kind of data across any industry reach out to:

Jay Grieve (jpgrieve551@gmail.com)

Aarjav Jain (aarjav02@gmail.com)

Dataset Description

This dataset contains synchronized ego and top-down video pairs capturing skilled workers performing diamond jewelry manufacturing tasks.

Job Types

Job Type	Description	Workers
setting	Diamond setting work	worker-001, worker-002, worker-003, worker-004, worker-005
wax_pulling	Wax pulling process	worker-006, worker-007
wax_setting	Wax setting process	worker-008

Data Structure

diamond_factory/
├── setting/
│   ├── worker-001/
│   │   ├── ego_001.mp4
│   │   ├── top_001.mp4
│   │   ├── ego_002.mp4
│   │   └── top_002.mp4
│   └── ...
├── wax_pulling/
│   └── ...
└── wax_setting/
    └── ...

Statistics

• Total videos: 83
• Total size: 883.01 GB
• Video pairs: 41
• Job types: 3 (setting, wax_pulling, wax_setting)
• Workers: 8

Video Types

• ego: Egocentric (first-person) view from worker's perspective
• top: Top-down overhead view of the workspace

Usage

from datasets import load_dataset

# Load the dataset (streams from S3)
ds = load_dataset('ThisIsPepper/EgocentricSpecialties', streaming=True)

for sample in ds['train']:
    print(sample['video_path'])
    print(sample['job_type'])
    print(sample['worker'])
    print(sample['video_type'])

License

Apache 2.0