ConductScienceConductScience
ToolsConductScience tool
ConductVisionFree in-browser calculator

Pose Tracking Config Wizard.

Plan DeepLabCut pose tracking projects with species presets, training time estimates, and auto-generated config.yaml files.

Run calculator Request a Quote
PrivateData stays in your browser
LiveNo sign-up required
Validated2026-04-08
CitableMethods and citation included

Calculator

Results update in place

Try it out

Load example Pose Tracking Config Wizard data to see the full workflow

Species Preset

When to use

  • Plan a new DeepLabCut or similar pose tracking project before recording
  • Estimate labeling workload and timeline for grant budgeting or lab scheduling
  • Generate a DLC-compatible config.yaml with species-appropriate body part and skeleton definitions
  • Compare GPU training time estimates for hardware planning
  • Validate video specifications against pose tracking best practices

Do not use for

  • For real-time pose tracking — this is a planning tool, not a tracker
  • As a substitute for DLC’s own frame extraction and labeling GUI
  • For multi-animal tracking estimates — maDLC has different labeling requirements

Label diverse behavioral states, not just walking

DLC accuracy depends on training set diversity. If your animal rears, grooms, freezes, or turns, ensure those poses are represented. Use DLC’s k-means frame extraction to maximize pose diversity.

Record at 30+ FPS for most rodent behaviors

Below 30 FPS, fast movements like grooming bouts or escape responses create motion blur that confuses the network. For Drosophila wing tracking, 120+ FPS may be necessary.

More body parts means more labeling time

Each additional body point adds 3 seconds per frame to labeling time. A 10-point rat skeleton with 500 frames takes ~4.2 hours. Budget labeling time realistically.

GPU training time scales linearly with labeled frames

Doubling labeled frames roughly doubles training time. An RTX 3090 handles 200 frames in ~2 hours and 1,000 frames in ~10 hours. Plan overnight training runs for large datasets.

1

Method

Recommended labeled frames computed as max(200, min(1000, totalFrames ×\times 0.005)) per DLC documentation. Labeling time estimated at 3 seconds per body point per frame. GPU training times scale linearly from empirical baselines: RTX 3090 ~2h, A100 ~45min, CPU ~24h for 200 frames. Species presets based on standard DLC body part configurations from Mathis et al. (2018) and Nath et al. (2019).

2

Validated

Last validated 2026-04-08. Calculations are designed for planning and documentation support; verify procurement decisions against manufacturer specifications or institutional SOPs.

3

How to cite

How to Cite

ConductScience Pose Tracking Config Wizard (v1.0). ConductScience, Inc. 2026. Available at: https://conductscience.com/tools/pose-tracking-config-wizard

Mathis A et al. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci. 2018;21(9):1281–1289.

Nath T et al. Using DeepLabCut for 3D markerless pose estimation across species and behaviors. Nat Protoc. 2019;14(7):2152–2176.

Pose Tracking Project Planning Fundamentals

DeepLabCut (DLC) uses transfer learning from pretrained neural networks (ResNet, EfficientNet) to track user-defined body parts in video. Project planning involves three key decisions:

Body Part Selection: Choose anatomically distinct landmarks visible in all frames. More points increase labeling time linearly but improve kinematic reconstruction.
Frame Sampling: DLC recommends labeling 200–1,000 frames sampled uniformly across videos. The k-means algorithm in DLC selects maximally diverse frames. Over-labeling beyond 1,000 frames rarely improves accuracy.
Training Configuration: DLC uses data augmentation (10× by default) including rotation, scaling, and contrast changes. Training converges in 200k–500k iterations, taking 2–5 hours on a modern GPU.

Common Pitfalls in Pose Tracking Setup

Several factors degrade pose tracking quality beyond frame count:

• Low frame rate: Below 30 FPS, fast movements create motion blur and missing poses. Grooming and rearing in rodents require 60+ FPS • Inconsistent lighting: Shadows and reflections shift apparent body part positions. Use diffuse, uniform illumination • Labeling inconsistency: The single largest source of DLC error. Establish labeling conventions before starting and cross-validate between labelers • Too few diverse poses: If training frames only show walking, DLC will fail on rearing, grooming, or turning. Sample frames from all behavioral states • Skeleton topology errors: Incorrect edge connections produce crossed wireframes. Verify skeleton topology before training • Over-reliance on augmentation: Augmentation helps but cannot compensate for fundamentally unrepresentative training sets

Frequently asked

325
Free tools
1,200+
Institutions
100%
Client-side
0
Uploads required

Next steps

Equipment

Video Tracking Systems

Camera systems and software for animal tracking

Browse Equipment

ConductVision

AI-powered behavioral analysis platform

Browse Equipment

Behavioral Mazes

Mazes and arenas for behavioral testing

Browse Quote

Need help?

Get a quote on equipment or talk to the ConductScience team.

Request quote