# j96w/DenseFusion

"DenseFusion: 6D Object Pose Estimation by Iterative Dense Fusion" code repository

Repository: https://github.com/j96w/DenseFusion
Canonical: https://ross.abutalabs.com/products/densefusion
Homepage: https://sites.google.com/view/densefusion
Language: Python
License: MIT
License Family: permissive
Last push: 2023-01-02T20:27:09+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2791, "days_push": 1339, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1276, forks 308 (observed 2026-08-28T04:04:13.114494+00:00)

## What it is
DenseFusion is the official PyTorch implementation of the paper '6D Object Pose Estimation by Iterative Dense Fusion', which estimates the 6D pose of known objects from RGB-D images using a dense pixel-wise feature fusion network with iterative refinement. It includes training and evaluation code for the YCB-Video and LineMOD datasets plus trained checkpoints.

## Use cases
- estimate 6D pose of objects from RGB-D images
- train a pose estimation model on YCB-Video dataset
- evaluate pose estimation on LineMOD
- run pose estimation for robot grasping research
- reproduce DenseFusion paper results
- fine-tune pose estimation on my own dataset

## When to choose
- you need RGB-D based 6D object pose estimation for robotics or research
- you want to reproduce or build on the DenseFusion paper
- you need trained checkpoints for YCB-Video or LineMOD benchmarks

## When to avoid
- you need actively maintained code with modern PyTorch versions
- you need CPU-only training, which is extremely slow
- you need a production-ready pose estimation pipeline with segmentation included
- you need Python 3.10+ or recent CUDA support out of the box

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, computer-vision, image-processing
- domain: computer-vision, robotics, deep-learning, machine-learning
- platform: python
- tags: pose-estimation, rgbd, pytorch, 6d-pose, research-code, object-pose, linemod, ycb-video, linux, gpu, cuda

## Member repositories
- j96w/DenseFusion (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:13.114494+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T05:02:57.150451+00:00, confidence not recorded.
  - readme: https://github.com/j96w/DenseFusion (fetched 2026-08-28T04:04:13.114494+00:00, sha 3c3a03dd707d)
  - homepage: https://sites.google.com/view/densefusion (fetched 2026-08-29T12:14:01.044726+00:00, sha 09b2dfe3fbef)
- Data as of 2026-08-30T08:39:29.467469+00:00.