Back to Templates

Train YOLO detector

Train a YOLO detector on your own labeled images and score it on a test part it never saw: split, train, predict, evaluate in one graph.

, 6 nodes, 4 connections, 8 variables

Get this flow
Read-only preview

About this template

Train a YOLO detector on your own labeled images and get its accuracy measured on images it never saw during training.

Overview

  • Model: YOLO detector (Ultralytics family) - fine-tuned here from a pre-trained checkpoint you pick
  • Task: Object detection
  • Input: A labeled image dataset with boxes and classes. No split needed: the template makes one.
  • Pipeline: split into train / validation / test -> train -> predict on the test part -> evaluate
  • Output: A trained detector registered as a model, a training report, and a full evaluation report on the held-out test part (mAP, AP50, AP75, precision, recall, F1, per-class breakdown, error analysis, prediction previews)

When you need it

You have labeled images and want a YOLO model that knows your classes, with a number you can defend: the score comes from a test part that neither the training nor the choice of the best checkpoint ever touched. Run it first to get a baseline, then re-run it on the same template as your labels grow. If you would rather compare detector families on your data, run the sibling templates for DEIM and EdgeCrafter and read the three reports side by side.

What you need

A labeled image dataset in OvalBee, every object boxed and classed. If it is not in OvalBee yet, bring it in with one of the import templates first. In the setup you pick the dataset and the pre-trained YOLO model to start from. Classes come from the dataset's own labels.

How it works

Step What happens
Split Your dataset is divided 70 / 15 / 15 into training, validation and test images (the ratios can be changed)
Train The detector is fine-tuned on the training images; the validation images are scored along the way to pick the best checkpoint. The template sets 30 epochs at 640 px and early stopping with patience 20; change them in the training settings
Predict The trained detector runs on the test images only, keeping every box above a low confidence of 0.05 so the evaluation sees the whole score range
Evaluate The predictions are scored against the test labels

What a run gives you

The trained model, ready to use in other OvalBee templates, and two reports. The training report states what the run was given and what it cost. The evaluation report gives mAP (50:95), AP50, AP75, precision, recall and F1 at an automatically chosen confidence, a per-class table, a breakdown of what kind of mistakes the model makes, a confusion matrix, a precision-recall curve, a calibration view, and the hardest images next to their labels. It also lists suspected annotation issues: confident predictions where nothing is labeled are often missing labels, not model errors.

For scale: on a PCB component dataset (9 classes, about 1,400 images) a 30-epoch run took about four minutes of training on one GPU and scored mAP 0.77 on its test part. This is one run, with one random split; your dataset will land elsewhere.

Because the whole chain is one OvalBee graph, re-running it after new labels arrive repeats the same split, training and scoring, so two runs are comparable. Each run draws a new random split, so compare runs on the same dataset version and expect some noise. If two trainings share one GPU they can run out of memory; run them one after another.