# AILab-CVC/UniRepLKNet

[CVPR 2024 & TPAMI 2025] UniRepLKNet

Repository: https://github.com/AILab-CVC/UniRepLKNet
Canonical: https://ross.abutalabs.com/products/unireplknet
Homepage: https://arxiv.org/abs/2311.15599
Language: Python
License: Apache-2.0
License Family: permissive
Topics: architecture, artificial-intelligence, convolutional-neural-networks, deep-learning, multimodal-learning
Last push: 2025-08-10T05:11:26+00:00

## Health v2 (maintenance only)
Score: 43/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 36, release rhythm 35, longevity 72
- inputs: {"age_days": 1013, "days_push": 388, "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 1072, forks 63 (observed 2026-08-28T04:03:28.636294+00:00)

## What it is
UniRepLKNet is a large-kernel ConvNet architecture (CVPR 2024, TPAMI 2025) that provides universal perception across image, audio, video, point cloud, and time-series modalities with a unified architecture. It ships PyTorch implementations and pretrained models achieving state-of-the-art ImageNet, COCO, ADE20K, audio recognition, and weather forecasting results.

## Use cases
- train a large-kernel CNN for image classification on ImageNet
- use one unified architecture for audio, video, and image recognition
- run time-series forecasting like global temperature and wind speed prediction
- get pretrained backbone weights for object detection or semantic segmentation
- compare ConvNet vs transformer performance across multiple modalities

## When to choose
- you need a fast, accurate ConvNet backbone for vision tasks like detection or segmentation
- you want a single architecture that handles multiple modalities without transformers
- you need state-of-the-art large-kernel ConvNet results with pretrained weights

## When to avoid
- you need a general-purpose deep learning framework rather than a specific model
- your project requires transformer-based multimodal models
- you work outside PyTorch or lack GPU resources

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, deep-learning, image-processing, audio-processing, video-processing, nlp
- domain: deep-learning, computer-vision, artificial-intelligence, image-processing
- platform: python, cross-platform
- tags: convnet, large-kernel, structural-reparameterization, multimodal, image-classification, time-series-forecasting, cvpr-2024, pretrained-models, audio, video, gpu

## Member repositories
- AILab-CVC/UniRepLKNet (main) score 43

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:28.636294+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 2026-08-30T06:53:39.346582+00:00, confidence not recorded.
  - readme: https://github.com/AILab-CVC/UniRepLKNet (fetched 2026-08-28T04:03:28.636294+00:00, sha c13a5018f38e)
  - homepage: https://arxiv.org/abs/2311.15599 (fetched 2026-08-29T12:55:56.735509+00:00, sha 408fe5813bed)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T12:55:56.737988+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T12:55:56.741344+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T12:55:56.743114+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T12:55:56.739627+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.