Home/Computer Vision/hold

hold

zc-alexfan/hold

Method for joint reconstruction of articulated hands and objects from monocular videos

GitHub Website

GraphCanon updated 3w · GitHub synced 3w

489 stars16 forksLast push 5mo Python MIT

Decision brief

HOLD for monocular video analysis of hand-object interactions without prior object models.

Good fit when

When no prior knowledge of object shapes is available but joint 3D reconstruction of hands manipulating objects from single-view videos is needed
For research projects requiring category-agnostic bimanual interaction reconstruction from custom captured sequences

Avoid when

If there are pre-scanned object models that could enhance accuracy beyond self-reconstruction capabilities
In scenarios where the computational resources for preprocessing and training on custom datasets are insufficient

Observed Jul 17, 2026 · Source: enrich:decision_facts

hold alternatives Trust report

Verify the decision

Maintenance and security

Full trust report

Maintenance: Slowing (143d since push); As of 3w
Provenance: Not a fork · Personal account; As of 3w
Security (OSV): 9 low (9 low); As of 1mo

Public GitHub metadata and optional OSV scans. Signals, not a guarantee. Trust methodology.

Install

pip install hold

PyPI

Similar tools

Same-category neighbours. No typed graph edges are catalogued for this tool yet.

transformers

Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models

164k★

PaddleOCR

A powerful, lightweight OCR toolkit to convert images and PDFs into structured data

88k★

stable-diffusion

A latent text-to-image diffusion model

73k★

Evidence and technical details

Sourced facts, taxonomy, compatibility claims, README excerpt, and machine-readable endpoints.

Overview

The HOLD method enables the 3D reconstruction of articulated hand-object interactions from single-view video inputs without needing prior object models or labeled 3D training data.

Capability facts

Languages: python
Source: github.language · Aug 1, 2026

README

Getting started

Get a copy of the code:

git clone https://github.com/zc-alexfan/hold.git
cd hold; git submodule update --init --recursive

Setup environments
- Follow the instructions here: docs/setup.md.
- You may skip external dependencies for now.
Train on a preprocessed sequence
- Start with one of our preprocessed in-the-wild sequences, such as hold_bottle1_itw.
- Familiarize yourself with the usage guidelines in docs/usage.md for this preprocessed sequence.
- This will enable you to train, render HOLD, and experiment with our interactive viewer.
- At this stage, you can also explore the HOLD code in the ./code directory.
Set up external dependencies and process custom videos
- After understanding the initial tools, set up the "external dependencies" as outlined in docs/setup.md.
- Preprocess the images from the hold_bottle1_itw sequence by following the instructions in docs/custom.md.
- Train on this sequence to learn how to build a custom dataset.
- You can capture your own custom video and reconstruct it in 3D at this point.
- Most preprocessing artifact files are documented in docs/data_doc.md, which you can use as a reference.
Two-hand setting: Bimanual category-agnostic reconstruction
- At this point, you can preprocess and train on a custom single-hand sequence.
- Now you can take on the bimanual category-agnostic reconstruction challenge!
- Following the instruction in docs/arctic.md to reconstruct two-hand manipulation of ARCTIC sequences.

For agents

This page has a .md twin and JSON over the API.

/tools/zc-alexfan-hold.md GET /api/graphcanon/tools/zc-alexfan-hold GET /api/graphcanon/graph?tool=zc-alexfan-hold