Deeplabcut

What it does

When your agent uses it

• The user needs animal pose estimation
• Behavior tracking
• Keypoint detection in videos
• Mentions DeepLabCut/DLC/SuperAnimal

Example prompts

• “/deeplabcut”

Requirements

• Python 3

Workflow steps

6 steps, taken from the first numbered list in SKILL.md.

1. 1Wrong PyTorch version: Always install PyTorch BEFORE deeplabcut. Check pytorch.org for the correct CUDA version.
2. 2GPU out of memory: Reduce batch size (default 8 → 4 or 2) in pose_cfg.yaml.
3. 3Missing GUI dependencies: label_frames requires pip install "deeplabcut[gui]". On headless servers, use X11 forwarding or label locally.
4. 4Video codec issues: If create_labeled_video fails, try converting to .mp4 (H.264) or .avi first.
5. 5maDLC detector: Multi-animal mode requires a detection model (fasterrcnn_resnet50_fpn_v2 is the default). Single-animal can work with…
6. 6SuperAnimal scale: Adjust scale_list parameter for SuperAnimal — try [200, 300, 400] first; smaller animals may need higher values.

What it can do on your machine

Read from SKILL.md and the folder at commit 93f6855. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • pip
  • conda
  • python

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  Links to these hosts (documentation or services it may open):

  • pytorch.org

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Deeplabcut loads about 1.7k tokens when it runs, and up to ~5.9k if it reads all its reference files. Until then it costs about 100 tokens; SKILL.md has 448 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~100

When it runs · the whole SKILL.md, loaded when a task matches

~1.7k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~5.9k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from NeuroAIHub/BrainPilot at commit 93f6855, republished under its AGPL-3.0 licence (© NeuroAIHub). 448 words, ~1,722 tokens.

name

deeplabcut

description

Toolbox for markerless animal pose estimation with DeepLabCut. Covers single/multi-animal tracking, SuperAnimal pretrained models, 2D/3D pose estimation, keypoint labeling GUI, model training/evaluation, video analysis, and behavioral quantification. Use when the user needs animal pose estimation, behavior tracking, keypoint detection in videos, or mentions DeepLabCut/DLC/SuperAnimal.

version

1.0.0

authors

Claude (AI-assisted)

review_status

ai-generated

source

https://github.com/DeepLabCut/DeepLabCut

What does the user need?
├── No labeled data, just want to track animals → SuperAnimal (video_inference_superanimal)
├── Single animal, have labeled data → Standard single-animal pipeline
├── Multiple animals interacting → maDLC (multi-animal pipeline)
├── 3D pose reconstruction → 3D pipeline (calibrate_cameras + triangulate)
└── Just post-process results → filterpredictions, analyzeskeleton

# Minimal (headless, no GUI):
pip install deeplabcut

# With GUI (label_frames, refine_labels, SkeletonBuilder):
pip install "deeplabcut[gui]"

# PyTorch must be installed separately:
pip install torch torchvision
# Or for GPU (check pytorch.org for your CUDA version):
conda install pytorch cudatoolkit=11.3 -c pytorch

import deeplabcut as dlc

# 1. Create project
config_path = dlc.create_new_project(
    "ProjectName", "ExperimenterName", ["/path/to/video.mp4"],
    working_directory="/path/to/projects"
)

# 2. Extract frames for labeling
dlc.extract_frames(config_path, mode="automatic", algo="kmeans", crop=True)

# 3. USER labels frames manually in GUI
# dlc.label_frames(config_path)  # launches the labeling GUI

# 4. Create training dataset from labeled frames
dlc.create_training_dataset(config_path, net_type="resnet_50")

# 5. Train the network
dlc.train_network(config_path, maxiters=100000, saveiters=5000)

# 6. Evaluate
dlc.evaluate_network(config_path, plotting=True)

# 7. Analyze videos (predict poses)
dlc.analyze_videos(config_path, ["/path/to/video.mp4"], videotype=".mp4")

# 8. Create labeled videos (overlay predictions)
dlc.create_labeled_video(config_path, ["/path/to/video.mp4"])

# 9. Export results to CSV
dlc.analyze_videos_converth5_to_csv("/path/to/videoDLC_resnet50_ProjectNameJul9")

© NeuroAIHub, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 5 other files (references) in packages/skills/skills/16_Animal_Behavior/deeplabcut of NeuroAIHub/BrainPilot.

• SKILL.md
• references/3d-pose.md
• references/maDLC.md
• references/modelzoo.md
• references/standard-pipeline.md
• references/utilities.md

Open the folder on GitHubat commit 93f6855