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

Queryinst

Queryinst is an open-source Python computer-vision research implementation of Instances as Queries, a query-based method for instance segmentation. Its repository includes training and inference code, configurations and pretrained weights, and is built on MMDetection with foundations in Sparse R-CNN and DETR.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Queryinst — 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 Queryinst when you need a research implementation and documented MMDetection workflow for training or evaluating instance-segmentation models. The repository's explicit training and evaluation examples use COCO; its video-segmentation context is described through QueryTrack and a demo.

Who it is for

  • Computer-vision researchers studying query-based instance segmentation.
  • Developers training or evaluating instance-segmentation models through MMDetection.

Documented uses

  • Train QueryInst using a provided COCO configuration.
  • Evaluate bounding-box and segmentation-mask predictions on COCO validation data.
  • Study the implementation, configurations and pretrained weights.
  • Explore video instance segmentation through the related QueryTrack work described in the repository.
Before you choose

Requirements & limits

The repository is public and open source under the MIT license.

  • The README directs users to MMDetection's installation guide and does not give a complete environment specification.
  • Training requires COCO data; evaluation also requires a checkpoint.
  • The repository recommends the newer QueryInst implementation included in MMDetection, while stating that this repository will also be maintained.

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

Train and evaluate QueryInst on COCO with one GPU

Run the README’s single-GPU COCO training and validation commands after setup and dataset preparation.

Before you start

  • QueryInst installed in development mode and COCO linked at data/coco, as described in the setup guide.
  • The QueryInst repository files and its listed COCO configuration.

Steps

  1. From the QueryInst project directory, train with the command below.

    Documented command
    python tools/train.py configs/queryinst/queryinst_r50_fpn_1x_coco.py

  2. Test on the COCO validation set with the command below, substituting the checkpoint path for PATH/TO/CKPT.pth.

    Documented command
    python tools/test.py configs/queryinst/queryinst_r50_fpn_1x_coco.py PATH/TO/CKPT.pth --eval bbox segm

Expected outcome

The documented commands start single-GPU training and run bbox and segmentation evaluation on COCO val using the specified checkpoint.

Guide limitations

  • The commands and configuration describe COCO; the README marks configurations for Cityscapes and YouTube-VIS as TODO.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
hustvl/QueryInst · Official homepage title · retained in its source language
How we review sourcesSuggest a correction