Skip to content

Dataset · cvml-nus

12-camera toy assembly and disassembly videos

cvml-nus/assembly101

Multi-view video recordings of toy vehicle assembly with action and error annotations.

What's inside

The corpus includes 4,321 video recordings of people assembling and disassembling 101 toy vehicle models without strict instructions, making natural errors and correcting them. Filming was conducted simultaneously from 8 static and 4 egocentric cameras. Text annotations in English are attached to the videos: over 100 thousand coarse and 1 million fine-grained action labels.

What you can do with it

  • Training video models for step-by-step action and error recognition
  • Developing AI assistants for real-time assembly monitoring and instruction
  • Research in multi-view computer vision and robotics
Languages
English
Size
up to 1K rows
Downloads this month
121.4K
Likes
38
Access
on request on Hugging Face

How to load it

from datasets import load_dataset

ds = load_dataset("cvml-nus/assembly101")
12-camera toy assembly and disassembly videos: dataset cvml-nus/assembly101 | NeyroLearn