Dataset labeling

Mirilux

AI-powered auto-labeling for building YOLO datasets. Import frames, run ensemble detection, review and correct labels, then export a dataset you can train on immediately.

Buy Mirilux Join Discord
YOLO datasetsEnsemble ONNXReview + export
Mirilux interface showing AI-assisted labeling

How it works

A workflow you can finish.

01

Load

Bring in image folders or extract frames from video. YouTube import pulls footage directly into your workspace.

YouTube import view
YouTube import
02

Auto-label

Run ensemble detection on batches to generate initial labels across every frame in your dataset.

Ensemble detection view
Ensemble detection
03

Review

Correct edge cases quickly with a verification-first UI. Manage classes and adjust bounding boxes before export.

Class management view
Class management
04

Export

Write YOLO-format labels and dataset splits for training. Optional built-in training pipeline with ONNX export.

Training workflow view
Built-in training

Formats

From frames to a YOLO dataset.

Input
Image folders, extracted frames, YouTube import
Labeling
Ensemble ONNX detection + manual review
Export
YOLO-format labels + dataset splits
Training
Optional built-in training pipeline + ONNX export

Ready to build a dataset faster?

Get Mirilux and ship labels that are consistent enough to train on.

Buy Mirilux