Synchronized video · vision models · live graph · teaching visuals
Play the source while hands, motion, attention, and the process graph appear on the same clock. Inspect what truly ran, what did not, and how source frames can become focused teaching visuals.
Nodes
Relations
Timecoded overlays
Real model artifacts
01 · VIDEO GROUNDING
Choose a process or graph node to seek its evidence moment. Every visible layer declares its model, provenance, and review state. Badges report retained source-time coverage—not the share of frames processed.
AI model result moments
74 model outputs03 · MODEL OUTPUT LEDGER
See the runtime truth: which model actually ran, how many artifacts it produced, and whether those outputs can be published.
CUDA runtime and an approved SAM 3.1 checkpoint are required; neither is available in this local build environment.
The official tracker environment/checkpoint has not been installed, and this runtime has no CUDA accelerator.
The official tracker environment/checkpoint has not been installed, and this runtime has no CUDA accelerator.
DINOv3 is conditional on a demonstrated re-identification failure; its approved weights/runtime are not present locally.
Landmarks are analysis previews until a human accepts coverage and identity for each clip; coverage alone cannot validate instructional meaning.
hand-clip-01.jsonA local pose-only screening found one strong result among four manually bounded bare-hand crops and no strong result among four glove crops; one glove skeleton was visibly misleading. Keep this challenger unavailable for publication until detector, full-clip temporal, source-hash and license gates pass.
Selected 3D-mesh challenger for grip geometry and partial occlusion. It requires an external CUDA run plus approved HaMeR checkpoint and MANO-model terms; no mesh output has been produced in this build.
Technically relevant for temporally consistent 4D interacting-hand motion, but the official code and models are restricted to non-commercial research and also depend on MANO. It is not admitted as a Denso demo dependency.
No repeated domain-labelled detection task has been defined; generic boxes would add visual noise without instructional evidence.
Reviewer- or VLM-proposed keyframes only; masks are not yet propagated through the full clip and remain review-pending.
keyframes.jsonclamp-lower-glove-mask.pngclamp-machine-chuck-mask.pngclamp-upper-glove-mask.pngclamp-valve-stem-mask.pnggrind-glove-mask.pnggrind-machine-chuck-mask.pnggrind-valve-stem-mask.pngReviewer- or VLM-proposed bounded clips only. Start/middle/end identity and drift checks remain review-pending; masks are suppressed outside generated windows.
48000-58000-4fps.json50000-58000-4fps.json90000-100000-4fps.json04 · SOURCE-TO-TEACHING VISUAL
Compare the source still, structure emphasis, and a teaching composition. Missing masks or generated images remain visibly missing rather than being simulated as model output.
05 · FULL GRAPH EXPLORER
Step outside the live view to inspect the source-specific process, attention, judgment, and evidence layers. The normalized learning structure continues in the Skill map.
Continue with this source
Switching views preserves the same source hash across instruction, analysis, skill structure, evidence, and human review.