# NVlabs/DG-Net

:couple: Joint Discriminative and Generative Learning for Person Re-identification. CVPR'19 (Oral) :couple:

Repository: https://github.com/NVlabs/DG-Net
Canonical: https://ross.abutalabs.com/products/dg-net
Homepage: https://www.zdzheng.xyz/publication/Joint-di2019
Language: Python
License: NOASSERTION
License Family: other
Topics: re-identification, person-reidentification, person-reid, image-retrieval, image-search, open-reid, pytorch, apex, market-1501, dukemtmc-reid, msmt17, cuhk-np, dg-net
Last push: 2023-07-09T10:24:47+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": 2623, "days_push": 1151, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1298, forks 226 (observed 2026-08-28T04:04:17.238060+00:00)

## What it is
DG-Net is a PyTorch implementation of the CVPR 2019 (Oral) paper 'Joint Discriminative and Generative Learning for Person Re-identification'. It jointly trains discriminative re-id feature learning with generative appearance swapping to improve robustness across camera views.

## Use cases
- train a person re-identification model
- search for a person across camera images
- run image retrieval on Market-1501 or DukeMTMC-ReID
- generate person images with swapped clothing appearance
- evaluate re-id models with multi-query and random erasing
- reproduce CVPR 2019 re-id paper results

## When to choose
- you need a proven person re-id baseline in PyTorch
- you want to combine generative data augmentation with re-id training
- you work with standard re-id benchmarks like Market-1501, MSMT17, or CUHK-NP
- you need multi-GPU training with optional fp16 via APEX

## When to avoid
- you need a commercially licensed model (CC BY-NC-SA 4.0)
- you need a maintained production system rather than research code
- you work outside person re-identification
- you cannot meet the 10-15GB GPU memory requirement

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, computer-vision, image-processing, search-engine
- domain: computer-vision, machine-learning, deep-learning, image-processing
- platform: python
- tags: person-reidentification, pytorch, generative-models, image-retrieval, research-code, cvpr-2019, non-commercial-license, search, linux, gpu

## Member repositories
- NVlabs/DG-Net (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:17.238060+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 2026-08-30T04:53:48.374623+00:00, confidence not recorded.
  - readme: https://github.com/NVlabs/DG-Net (fetched 2026-08-28T04:04:17.238060+00:00, sha 47ab9e09269b)
  - homepage: https://www.zdzheng.xyz/publication/Joint-di2019 (fetched 2026-08-29T12:10:12.133622+00:00, sha b1444e7e123f)
- Data as of 2026-08-30T08:39:29.467469+00:00.