Shotwell
Blog Demo Talk to us Upload videos now · 24 hr turnaround

Annotations for robotics data

Shotwell provides fast, accurate and dense annotations for robotics training data. We segment long videos into discrete steps and detect failures in each step.

Upload videos now Talk to us
episode0431 · fold_clothing
source1278×720 · 19.1s
step1/8
subtask
rubric
verdictPASS
Subtasks
Pick up✓
0:00
Trusted by
Weave Robotics Ultra Frontier Machines Parametric

"Shotwell annotations improved our training data visibility to allow every QA reviewer to review 6x more operators."

0:10.00
Precise timestamps Sub-second subtask boundaries with SOP validation, for stronger policy training and enabling RL techniques.
where  subtask="Set Yellow"
and    result="fail"
return episode_id, t0, t1
Dataset observability Failure detection and attribute descriptions per subtask. Manual review becomes a database query.
↑ uploadedmon 18:02
↓ returnedtue 17:31
23h 29m
24-hour feedback Operator notes returned while runs are still fresh.

Let's talk about your data: hello@shotwell.ai

Shotwell