RectLabel
An offline image annotation tool for object detection and segmentation.
To download RectLabel apps.
Post the problem to our Github issues.
Have questions? Send an email to support@rectlabel.com.
How to Train a YOLO26 Object Detection Model with Custom Data
We will show you how to train a YOLO26 detection model with your images and annotations and export to a Core ML model which can be used for auto labeling on RectLabel.
We recommend working through this blog post side-by-side with the YOLO26 Object Detection Colab notebook.
Install YOLO26.
pip install -q ultralytics
Download training images and annotations. You can use these or replace them with your own data.
mkdir datasets
cd datasets
wget -q https://huggingface.co/datasets/rectlabel/datasets/resolve/main/converse_vans_detection.zip
unzip -q converse_vans_detection.zip
cd ..
Create a workspace folder and start training from the workspace folder. Make sure the datasets path in the yaml file.
mkdir workspace
cd workspace
mv ../datasets/converse_vans_detection/converse_vans_detection.yaml .
yolo task=detect mode=train model=yolo26n.pt data=converse_vans_detection.yaml epochs=100 imgsz=640 plots=True
Move the best model to the current folder and export to a Core ML model.
mv runs/detect/train/weights/best.pt .
yolo export model=best.pt format=coreml
Now you can auto label using the Core ML model on RectLabel.