Datasets:
File size: 3,017 Bytes
6c1c833 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 | {
"field_id": "F005",
"field": "3d scene reconstruction",
"directions": [
{
"direction_id": "G0",
"name": "Surface recovery from visual and measured geometry",
"scope": "Recover accurate, continuous or extractable scene surfaces from geometric evidence provided by multiview images, depth observations, scans or point samples. This includes geometric constraints, surface representations, rendering-based optimization, and handling sparse or noisy measurements."
},
{
"direction_id": "G1",
"name": "Geometry inference with missing visual evidence",
"scope": "When views are scarce, only a single image of an object is available, or scene regions are occluded, use priors, generative guidance and geometric consistency to infer insufficiently observed 3D content. Outputs include renderable scenes, individual object shapes and completed scene geometry."
},
{
"direction_id": "G2",
"name": "Resource-efficient scene representations",
"scope": "Reduce the storage, training or runtime costs of neural scene reconstruction and rendering, while addressing the representation of large-scale scenes."
},
{
"direction_id": "G3",
"name": "Joint camera alignment and structure recovery",
"scope": "When camera poses or calibration are unreliable, recover consistent 3D structure while establishing or refining the spatial relationships among views."
},
{
"direction_id": "G4",
"name": "Continual dense scene mapping",
"scope": "Continuously update a dense scene representation as sensor observations arrive, handling the fusion of new and prior observations, pose revisions or retention of previously learned content."
},
{
"direction_id": "G5",
"name": "Reconstruction of changing scenes",
"scope": "Jointly recover 3D geometry or appearance that changes over time and the motion, deformation, articulation or temporal correspondence needed to explain those changes."
},
{
"direction_id": "G6",
"name": "Structured and interpretable scene reconstruction",
"scope": "Reconstruct not only overall geometry but also the spatial organization of a scene: semantically labeled 3D regions, distinguishable objects or parts, and explicit room or building structure."
},
{
"direction_id": "G7",
"name": "Reconstruction through explicit measurement models",
"scope": "For observations that ordinary geometry and pixel correspondences cannot adequately explain, explicitly model image-formation processes such as illumination, materials, reflection and refraction, or event-camera measurements of brightness changes, to recover scene geometry and recoverable appearance."
},
{
"direction_id": "G8",
"name": "Active observation for reconstruction",
"scope": "Decide which view to acquire or interaction to perform next to reveal missing geometry and improve an ongoing 3D reconstruction."
}
]
}
|