nAIve physics
Menu
nAIve physicsInternal / Experimental Workbench
Temporalv0.2.1
  1. 1Upload
  2. 2Verify metadata
  3. 3Annotate
  4. 4Export
  5. 5Next video

FREE TOOL / DATA WORKBENCH

Turn video into temporal data.

Convert manipulation video into human-verified temporal data.
Temporal behavior

Source: Human annotated

Frame index: Derived from timestamp × FPS

01 / VIDEO INPUT

Choose a manipulation video.

The selected file stays in this browser session. It is not automatically added to the nAIve physics commercial dataset.

02 / AUTO CAPTURE

Source file

Awaiting video
File nameNo file selectedAUTO-DETECTED
File sizeNot availableNOT AVAILABLE
ResolutionAwaiting metadataAUTO-DETECTED
DurationAwaiting metadataAUTO-DETECTED
MIME typeNot reportedAUTO-DETECTED
FPSFPS NOT DETECTEDMANUAL REQUIRED
Episode IDEpisode ID not detectedMANUAL REQUIRED

03 / VERIFY METADATA

Batch metadata

Locked defaults apply to the next video

FPS REQUIRED for frame indices. Time-based annotation remains available.

No video loadedUpload a local video or try the sample episode.
Current time0.00 s
Current frameFPS REQUIRED
MediaAwaiting metadata
EpisodeREQUIRED
Space Play / pause[ Mark start] Mark end

03 / BEHAVIOR TRACK

Timeline

00:00 — 00:00
Temporal coverage0.000 / 0.000 sec0.0%
Unannotated0.000 sec0 gaps
Overlap QA0NONE DETECTED
0.0 s0.0 s0.0 s0.0 s0.0 s

No gaps or overlaps detected.

04 / HUMAN VERIFICATION

Annotation controls

Experimental interaction labels are being evaluated and are not part of the published v0.2 ontology.

Start
Not marked
End
Not marked
Duration

06 / SESSION RECORDS

Annotation records

Annotations are not saved until exported.
TEMPORAL QA INCOMPLETESegments: 0Coverage: 0.0%Gaps: 0Overlaps: 0Episode: 0.000 sec
No temporal annotations yet.

Choose a video, mark a start and end point, then create the first human-verified behavior segment.

INTERNAL / EXPERIMENTAL WORKBENCH

Auto analysis

Independent browser-side layers. Failure in one module does not block temporal annotation.
Density
GARMENTnot seeded
IRONnot initialized

Load an episode to begin.

Garment segmentationnot started
Semantic keypointsnot started
Hand posenot started
Iron trackingnot started
Optical flownot started
Slip candidatesnot started
RGB analysisnot started
Depth estimatenot started
Estimated 3Dnot started

INTERPRETABLE TEMPORAL PROPOSALS

Temporal review

Automation proposes intervals. Only an explicit human acceptance adds a verified annotation.
EPISODE STATEinitial_state · terminal_state
HUMAN BEHAVIOURapproach · position · tension · iron_stroke · iron_hold · reposition · inspect · release
INTERACTIONanchor
FABRIC RESPONSEslip

Run Auto Analysis, seed the garment and iron where needed, then generate temporal candidates. Nothing is generated or accepted automatically.

SEMANTIC + TRACKING SEED

Source-pixel inspection

Capture a frame, choose an interaction, then click. MediaPipe loads only when analysis or garment segmentation is requested.

Mask: not seeded. Hands: 0 detections. Pose-support samples: 0. Tool: 0 tracked positions. Model proposals and tracked outputs remain reviewable estimates.

CONSOLIDATED REVIEW

Review required

0

No automatic review items yet. Run analysis to generate interpretable candidates.

RGB samples 0Mask tracks 0Slip candidates 0Accepted slip 0
INTERACTION CONTEXTAuthoritative analysis revision 0
Garment mask records0
Iron auto-track records0
Temporal contextUNANALYZED

EXPERIMENTAL OUTPUTS

Export only what exists

Canonical temporal CSV remains nine columns and excludes unaccepted candidates.

MediaPipe Tasks Vision 1.0.1; Hand Landmarker, Pose Landmarker, and Interactive Segmenter models load remotely from Google-hosted assets. Model URLs are recorded in provenance: hand, pose, segmenter.

Scientific statusOptical flow is image motion, not calibrated physical velocity.Slip candidates are not verified slip until accepted by a human.RGB response is experimental: not ground truth, physical wrinkle height, or measured geometry.Depth not available in current deployment. Estimated 3D unavailable without depth.

Thermal analysis is not available from RGB video. It requires thermal / IR sensor data.