Guides¶

✏️ Labeling Instances

Placing, editing, selecting, and deleting instances and nodes.

🦴 Skeletons

Templates, drawing a skeleton on the frame, symmetries, editing nodes and edges.

🧭 Navigation

Frames, the seekbar, statistic graphs, marks, and multi-video projects.

💡 Suggestions

Choosing which frames are worth labeling, seven ways.

👥 Tracks

Identities across frames: assigning, transposing, propagating, cleaning up.

🎬 Videos

Adding videos, remote URLs, missing-path resolution, codecs and transcoding.

👁️ View & Display

Zoom, pan, colors, palettes, trails, node sizes, and what gets drawn.

🐍 Environment setup

Getting the Python backend in place with uv, no conda required.

🧠 Training

Model types, configuration, monitoring, and metrics.

⚡ Inference

Running predictions and bringing them back into the project.

🌐 Remote Compute

Pairing a GPU worker and sending it training and inference jobs.

🔄 Import & Export

NWB, COCO, DeepLabCut, Analysis HDF5/CSV, packages, labeled clips.

🔀 Merging Projects

Combining two projects, with conflict review.

💾 Saving & Recovery

How saving differs by runtime, and the crash-recovery draft system.

📊 Instance Size Distribution

How big are your animals, really? Pick a crop size from data.

🔍 Label Quality Check

Automated checks for duplicates, left/right swaps, and chain-order errors.