# microsoft/Oscar

Oscar and VinVL

Repository: https://github.com/microsoft/Oscar
Canonical: https://ross.abutalabs.com/products/oscar
Language: Python
License: MIT
License Family: permissive
Topics: vision-and-language, pre-training, image-captioning, vqa, image-text-search, oscar, vinvl
Archived: true
Last push: 2023-08-28T01:34:59+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2302, "days_push": 1102, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1053, forks 249 (observed 2026-08-28T04:03:23.897603+00:00)

## What it is
Oscar is Microsoft's research code for object-semantics aligned cross-modal pre-training of vision-language models, with VinVL providing improved visual representations. It includes pretrained checkpoints and fine-tuning code for vision-language understanding and generation tasks.

## Use cases
- pretrain a vision-language model on image-text pairs
- fine-tune a model for visual question answering
- generate image captions with a pretrained model
- search images using text queries
- reproduce Oscar or VinVL paper results
- extract object-attribute features for V+L tasks

## When to choose
- you need proven state-of-the-art vision-language models like Oscar/VinVL
- you are doing research on VQA, image captioning, or image-text retrieval
- you want pretrained checkpoints for cross-modal pre-training experiments

## When to avoid
- you need a maintained production library - the repo is in maintenance mode with no active development
- you want modern multimodal LLMs - the authors point to LLaVA for instruction-tuned models
- you need a simple inference API rather than research training code

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, nlp, image-processing
- domain: artificial-intelligence, computer-vision, deep-learning
- platform: python
- tags: vision-and-language, pre-training, image-captioning, vqa, image-text-retrieval, multimodal, research-code, vinvl, natural-language-processing

## Member repositories
- microsoft/Oscar (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:23.897603+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 2026-08-30T06:59:22.363373+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/Oscar (fetched 2026-08-28T04:03:23.897603+00:00, sha 313b00c272de)
- Data as of 2026-08-30T08:39:29.467469+00:00.