docs/assets/charts/research_direction_extension_tasks.svg

<svg xmlns="http://www.w3.org/2000/svg" width="1420" height="920" viewBox="0 0 1420 920">
<rect width="1420" height="920" fill="#020502"/>
<rect x="28" y="28" width="1364" height="864" rx="18" fill="#050905" stroke="#ccffa0" stroke-opacity="0.24"/>
<text x="66" y="88" font-size="32" font-weight="760" fill="#f4f8ef">Ropedia Xperience-10M: four direction extension probes</text>
<text x="66" y="122" font-size="17" font-weight="500" fill="#a5afa2">Data-backed from the same 1,161-window public sample feature tensor; extension probes for later held-out studies.</text>
<rect x="66" y="166" width="620" height="160" rx="10" fill="#071207" stroke="#ccffa0" stroke-opacity="0.22"/>
<rect x="66" y="166" width="10" height="160" rx="5" fill="#ccffa0"/>
<circle cx="108" cy="206" r="24" fill="#ccffa0" opacity="0.14"/>
<text x="98" y="214" font-size="21" font-weight="760" fill="#ccffa0">A</text>
<text x="142" y="201" font-size="20" font-weight="760" fill="#f4f8ef">Body and Hand Motion Intensity</text>
<text x="142" y="228" font-size="13" font-weight="650" fill="#a5afa2">Human Modeling &amp; Motion Understanding</text>
<text x="142" y="260" font-size="16" font-weight="700" fill="#f4f8ef">Minimal: 0.7658 macro-F1</text>
<text x="366" y="260" font-size="16" font-weight="700" fill="#f4f8ef">Neural MLP: 0.8254 macro-F1</text>
<text x="142" y="291" font-size="13" font-weight="500" fill="#dce8d7">Binary label: high_motion or low_motion.</text>
<rect x="142" y="304" width="440" height="8" rx="4" fill="#ccffa0" opacity="0.14"/>
<rect x="142" y="304" width="337.0" height="8" rx="4" fill="#ccffa0" opacity="0.72"/>
<rect x="142" y="316" width="440" height="8" rx="4" fill="#ccffa0" opacity="0.14"/>
<rect x="142" y="316" width="363.2" height="8" rx="4" fill="#ffffff" opacity="0.78"/>
<rect x="730" y="166" width="620" height="160" rx="10" fill="#071207" stroke="#ccffa0" stroke-opacity="0.22"/>
<rect x="730" y="166" width="10" height="160" rx="5" fill="#7ae5c3"/>
<circle cx="772" cy="206" r="24" fill="#7ae5c3" opacity="0.14"/>
<text x="762" y="214" font-size="21" font-weight="760" fill="#7ae5c3">B</text>
<text x="806" y="201" font-size="20" font-weight="760" fill="#f4f8ef">Multi-View Consistency Retrieval</text>
<text x="806" y="228" font-size="13" font-weight="650" fill="#a5afa2">3D/4D Reconstruction &amp; Neural Rendering</text>
<text x="806" y="260" font-size="16" font-weight="700" fill="#f4f8ef">Minimal: 0.5534 MRR</text>
<text x="1030" y="260" font-size="16" font-weight="700" fill="#f4f8ef">Neural MLP: 0.3469 MRR</text>
<text x="806" y="291" font-size="13" font-weight="500" fill="#dce8d7">Ranked candidate windows; the correct synchronized view should rank near the top.</text>
<rect x="806" y="304" width="440" height="8" rx="4" fill="#ccffa0" opacity="0.14"/>
<rect x="806" y="304" width="243.5" height="8" rx="4" fill="#7ae5c3" opacity="0.72"/>
<rect x="806" y="316" width="440" height="8" rx="4" fill="#ccffa0" opacity="0.14"/>
<rect x="806" y="316" width="152.6" height="8" rx="4" fill="#ffffff" opacity="0.78"/>
<rect x="66" y="360" width="620" height="160" rx="10" fill="#071207" stroke="#ccffa0" stroke-opacity="0.22"/>
<rect x="66" y="360" width="10" height="160" rx="5" fill="#d8f4a5"/>
<circle cx="108" cy="400" r="24" fill="#d8f4a5" opacity="0.14"/>
<text x="98" y="408" font-size="21" font-weight="760" fill="#d8f4a5">C</text>
<text x="142" y="395" font-size="20" font-weight="760" fill="#f4f8ef">Action Phase Progress Estimation</text>
<text x="142" y="422" font-size="13" font-weight="650" fill="#a5afa2">Egocentric Vision &amp; Interaction</text>
<text x="142" y="454" font-size="16" font-weight="700" fill="#f4f8ef">Minimal: 0.3267 MAE</text>
<text x="366" y="454" font-size="16" font-weight="700" fill="#f4f8ef">Neural MLP: 0.3015 MAE</text>
<text x="142" y="485" font-size="13" font-weight="500" fill="#dce8d7">A scalar progress value between 0.0 and 1.0 for the current action segment.</text>
<rect x="142" y="498" width="440" height="8" rx="4" fill="#ccffa0" opacity="0.14"/>
<rect x="142" y="498" width="296.2" height="8" rx="4" fill="#d8f4a5" opacity="0.72"/>
<rect x="142" y="510" width="440" height="8" rx="4" fill="#ccffa0" opacity="0.14"/>
<rect x="142" y="510" width="307.3" height="8" rx="4" fill="#ffffff" opacity="0.78"/>
<rect x="730" y="360" width="620" height="160" rx="10" fill="#071207" stroke="#ccffa0" stroke-opacity="0.22"/>
<rect x="730" y="360" width="10" height="160" rx="5" fill="#9bdfff"/>
<circle cx="772" cy="400" r="24" fill="#9bdfff" opacity="0.14"/>
<text x="762" y="408" font-size="21" font-weight="760" fill="#9bdfff">D</text>
<text x="806" y="395" font-size="20" font-weight="760" fill="#f4f8ef">Short-Horizon Ego-Motion Forecasting</text>
<text x="806" y="422" font-size="13" font-weight="650" fill="#a5afa2">Scene Reconstruction &amp; World Modeling</text>
<text x="806" y="454" font-size="16" font-weight="700" fill="#f4f8ef">Minimal: 0.1700 MAE</text>
<text x="1030" y="454" font-size="16" font-weight="700" fill="#f4f8ef">Neural MLP: 0.0833 MAE</text>
<text x="806" y="485" font-size="13" font-weight="500" fill="#dce8d7">A future camera-translation delta vector.</text>
<rect x="806" y="498" width="440" height="8" rx="4" fill="#ccffa0" opacity="0.14"/>
<rect x="806" y="498" width="365.2" height="8" rx="4" fill="#9bdfff" opacity="0.72"/>
<rect x="806" y="510" width="440" height="8" rx="4" fill="#ccffa0" opacity="0.14"/>
<rect x="806" y="510" width="403.3" height="8" rx="4" fill="#ffffff" opacity="0.78"/>
<text x="66" y="570" font-size="24" font-weight="760" fill="#f4f8ef">How to read this</text>
<text x="66" y="604" font-size="16" font-weight="500" fill="#dce8d7">Each card adds one concrete task to a research direction using existing sample modalities.</text>
<text x="66" y="632" font-size="16" font-weight="500" fill="#dce8d7">Colored bar: minimal baseline normalized score. White bar: neural MLP normalized score. Lower-is-better MAE is shown as 1 - MAE for bar length only.</text>
<line x1="66" y1="675" x2="1354" y2="675" stroke="#ccffa0" stroke-opacity="0.18"/>
<text x="66" y="724" font-size="22" font-weight="760" fill="#f4f8ef">Implementation boundary</text>
<text x="66" y="758" font-size="16" font-weight="500" fill="#dce8d7">A: motion-energy proxy, not a full human body model. B: view-feature retrieval, not neural rendering.</text>
<text x="66" y="786" font-size="16" font-weight="500" fill="#dce8d7">C: phase-progress regression, not open-world intent. D: ego-motion forecast, not a persistent map.</text>
<text x="66" y="835" font-size="16" font-weight="700" fill="#f4f8ef">All metrics are computed from held-out chronological windows of the same public sample episode.</text>
</svg>