[general]
name=QModel Trainer
qgisMinimumVersion=3.40
qgisMaximumVersion=3.99
description=Train YOLO11 and Pytorch models directly in QGIS with real-time monitoring
version=0.9.0
author=Nextelia
email=hello@nextelia.fr

about=about=QModel Trainer is a QGIS plugin for deep learning model training. It supports SMP, Mask R-CNN, and YOLO11 models with data augmentation, learning rate finder, resume training, early stopping, HTML reports, and QMTP export. AI features require QGeoAI Server for PyTorch/CUDA acceleration. Supports both GPU and CPU training.

tracker=https://github.com/nextelia/qmodeltrainer/issues
repository=https://github.com/nextelia/qmodeltrainer
homepage=https://qgeoai.nextelia.fr

hasProcessingProvider=no

changelog=0.9.0 - Pre-release (January 2026)
    - SMP semantic segmentation with multiple backbones
    - Instance segmentation Mask R-CNN
    - YOLO11 support: detection, segmentation, and OBB tasks
    - Real-time training metrics
    - Data augmentation
    - Learning rate finder, optimizer and scheduler
    - Resume training from checkpoints
    - Early stopping with configurable patience
    - HTML training reports
    - QMTP model export format

tags=deep-learning,geoai,training,YOLO11,UNet,segmentation,detection,pytorch,CUDA,QGeoAI,geospatial-AI

category=Raster
icon=assets/images/icon.png

experimental=False
deprecated=False

server=False