Skip to content

Centific QC SDK

Send your robot-capture data and get back a quality report: a pass/fail verdict, a score out of 100, exactly what to fix, and a PDF you can share.

The grading runs on our servers, not on your computer. You upload a zip, and the report comes back to you.

Get started See the scripts

From zip to report in four steps

  • 1. Install

    Install the Python SDK from the beta package in one command.

    Install the SDK

  • 2. Get an API key

    Create your key on the dashboard and load it into your terminal.

    Get an API key

  • 3. Prepare your data

    One zip, one data type, up to 5 GB, plus one sentence describing the task.

    Prepare your data

  • 4. Run a script

    Copy a ready-made script, run it, and save your PDF report.

    Create a job

What it looks like

from centific import PhysicalAI

client = PhysicalAI(api_key="sk-cen-...", base_url="https://humanoidiq-dev.centific.com")

job = client.qc.create(dataset="episode_0000.zip", data_type="egocentric",
                       task="Folding laundry and placing it on a bed")
print(job.id)                 # your receipt

report = job.wait()           # grading runs on our side
print(report.verdict, report.score)
job.save_pdf("report.pdf")

What you get back

  • A verdict and a score

    PASS or FAIL, and a score from 0 to 100 across video quality, framing, tracking, movement and task completion.

  • What to fix

    A list of concrete fixes, most important first, and the score you would reach if you applied them.

  • A shareable PDF

    The same report as a document you can send to a colleague or customer, with stills from your capture.

  • Many episodes, one report

    A single zip can hold many episodes. You get one report with a row for each episode.

Python, Node.js or the command line

The guide uses Python throughout. The same calls exist in the Node.js SDK and a shell script.