SAM 3.1segmentation
CUDA runtime and an approved SAM 3.1 checkpoint are required; neither is available in this local build environment.
Physical-process ontology · v1
Turn analysis into reusable learning structure.
Transform the video analysis into a canonical graph connecting process, tools, workpieces, attention, assessment, and learning activities. Selecting a node returns from structure to its evidence moment.
67Objects
160Relations
19Evidence sources
0Structural errors
01 · Skill graph
Skill graph
Each view is a projection of the same data. Node IDs and evidence remain stable when the view changes.
Knowledge categories · node backgroundQuality
02 · Video overlays
Video overlays
GRD-01 grinding machine — capability overview0:00
03 · MODEL LEDGER
Executed visual models and admission decisions
The interface reports run status, artifacts and limits—not just model names. Capabilities that did not run are not presented as demo outputs.
Runtime unavailable
TAPNext++point tracking
The official tracker environment/checkpoint has not been installed, and this runtime has no CUDA accelerator.
CoTracker3point tracking
The official tracker environment/checkpoint has not been installed, and this runtime has no CUDA accelerator.
DINOv3correspondence
DINOv3 is conditional on a demonstrated re-identification failure; its approved weights/runtime are not present locally.
MediaPipe Hand Landmarkerhand pose · Baseline
Landmarks are analysis previews until a human accepts coverage and identity for each clip; coverage alone cannot validate instructional meaning.
RTMPose Handhand pose · Selected 2D challenger
A 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.
HaMeR 3D hand meshhand mesh · Selected 3D challenger
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.
Dyn-HaMRhand mesh · Research comparator only
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.
YOLOdetection
No repeated domain-labelled detection task has been defined; generic boxes would add visual noise without instructional evidence.
SAM 2.1 Hiera Tinysegmentation
Reviewer- or VLM-proposed keyframes only; masks are not yet propagated through the full clip and remain review-pending.
SAM 2.1 Hiera Tiny video predictorsegmentation
Reviewer- or VLM-proposed bounded clips only. Start/middle/end identity and drift checks remain review-pending; masks are suppressed outside generated windows.
Continue with this source
One source, five connected views
Switching views preserves the same source hash across instruction, analysis, skill structure, evidence, and human review.
01Learning guideLearn the demonstration as a visual procedure02Vision labInspect model outputs, live graph, and teaching visuals
03Skill mapExplore process, tools, judgment, and learningYou are here
04EvidenceReturn to speech, source text, and observations05Open questionsAnswer knowledge gaps and update the skill graph