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

Omdet

OmDet is an open-source Python and PyTorch implementation of OmDet-Turbo, a transformer-based open-vocabulary object-detection model that locates objects in images from text descriptions.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Omdet — 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 OmDet when an image-analysis workflow needs object detection guided by text descriptions rather than only a fixed set of categories, and you can work with the model code and checkpoints.

Who it is for

  • Computer-vision developers building text-prompted image detection features
  • Teams exploring zero-shot detection of object categories not fixed in advance
  • Developers who want a local inference demo, a documented server option, or an ONNX export path

Documented uses

  • Detect and locate objects in images based on text descriptions.
  • Prototype image analysis for categories that are not fixed in advance.
  • Generate annotated prediction images with the local demo.
  • Explore serving inference through the repository's documented API-server script.
  • Export the configured model to ONNX, with additional work for post-processing or variable input sizes.
Before you choose

Requirements & limits

The GitHub repository is public and specifies an Apache-2.0 license. Model links are provided for Hugging Face and ModelScope.

  • The repository supplies model code and downloadable artifacts; it does not establish a hosted API service.
  • The documented local demo and server both require the pretrained model and CLIP checkpoints.
  • The README's ONNX export example omits post-processing and fixes input sizes.
  • Performance figures in the README refer to specified model variants, datasets, and hardware.

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

Run local image inference

Use the repository's documented local demo to produce images with predicted object detections.

Before you start

  • OmDet-Turbo repository and its installation instructions
  • A pretrained OmDet model checkpoint
  • The CLIP checkpoint
  • A configured environment for OmDet-Turbo

Steps

  1. Follow the repository's installation instructions to set up the environment.

  2. Download the pretrained model and CLIP checkpoint.

  3. Create a folder named resources and place both downloaded model files in it.

  4. Run run_demo.py; the README says images with predicted results are saved in ./outputs.

Expected outcome

The demo saves images with predicted detections in the ./outputs folder.

Guide limitations

  • The README does not specify the exact command-line invocation or input-image configuration.
  • This guide describes the repository's documented steps, not a tested execution.
Documentation 1 ↗
Source-based practical guide

Export the configured model to ONNX

Follow the repository's documented steps to export an ONNX model.

Before you start

  • The OmDet-Turbo repository and its installation instructions
  • The OmDet-Turbo-Tiny configuration file

Steps

  1. In OmDet-Turbo_tiny_SWIN_T.yaml, replace OmDetV2Turbo with OmDetV2TurboInfer.

  2. Run export.py to produce omdet.onnx.

Expected outcome

The documented export produces an omdet.onnx file.

Guide limitations

  • The example ONNX model excludes post-processing and uses fixed input sizes; the README says these can be changed or extended as needed.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
om-ai-lab/OmDet · Official homepage title · retained in its source language
How we review sourcesSuggest a correction