OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Convmae

ConvMAE is a Python research implementation of a masked-convolution masked autoencoder for learning visual representations and applying them to computer-vision tasks.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Convmae — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use ConvMAE when you need a research codebase with separately documented workflows for masked-image pretraining, fine-tuning, detection or segmentation.

Who it is for

  • Computer-vision researchers exploring masked-image modeling and representation learning.
  • Developers adapting a pretrained visual backbone for object detection or semantic segmentation.

Documented uses

  • Pretrain a ConvMAE model on ImageNet-1K.
  • Fine-tune a pretrained model or perform linear probing.
  • Use a pretrained backbone with Mask R-CNN for object detection.
  • Use a pretrained backbone with UperNet for semantic segmentation.
Before you choose

Requirements & limits

The public repository is released under the MIT License.

  • The documented environment requires Linux, Python 3.7 or later, CUDA 10.2 or later and GCC 5 or later.
  • Detection and segmentation use separate task guides and named downstream frameworks: Mask R-CNN and UperNet.
  • The repository points to a separate project, VideoConvMAE, for video classification.
  • The README says the project name changed from ConvMAE to MCMAE.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Find the documented ConvMAE task workflow

Choose the project guide for pretraining, finetuning, object detection, or semantic segmentation.

Before you start

  • Linux
  • Python 3.7 or later
  • CUDA 10.2 or later
  • GCC 5 or later

Steps

  1. Open the ConvMAE repository README.

  2. For ImageNet pretraining, follow PRETRAIN.md.

  3. For pretrained-model finetuning or linear probing, follow FINETUNE.md.

  4. For object detection with a pretrained backbone on Mask R-CNN, follow DETECTION.md; for semantic segmentation with a pretrained backbone on UperNet, follow SEGMENTATION.md.

Expected outcome

The repository points to a task-specific guide for each listed workflow.

Guide limitations

  • The README identifies the task guides but does not include their detailed procedures here.
  • Video classification is directed to the separate VideoConvMAE repository.
Documentation 1 ↗Documentation 2 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
Alpha-VL/ConvMAE · Official homepage title · retained in its source language
How we review sourcesSuggest a correction