Automated reconstruction of smoothly joining 3D printed restorations to fix broken objects

Lamb, Nikolas; Banerjee, Sean; Banerjee, Natasha Kholgade

doi:10.1145/3328939.3329005

Cited by 12 publications

(5 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Currently the biggest challenge to progress in the field of fracture repair is the dearth of real data, with the largest existing dataset of real fractured objects with complete counterparts containing only 22 samples [LBB19]. The generalizability of our approach to real fractures may be limited by the simple structure of objects from ShapeNet, which are derived from computer aided design (CAD) models.…”

Section: Limitations and Future Workmentioning

confidence: 99%

“…Our fracturing approach also simulates the subset of possible fractures that can be generated using subtraction with a randomized geometric primitive, i. As part of ongoing work, we are actively collecting a dataset of real fractured objects with ground truth complete counterparts, going well beyond the object, material, and fracture-type sets in Lamb et al [LBB19]. We are generating fractures for common materials and fracture types, e.g., ceramic chipping, wood splintering, wood shearing along grains, and large fractures for ceramic, clay, hard plastic, and concrete.…”

Section: Limitations and Future Workmentioning

confidence: 99%

“…Our approach addresses the challenge that fractured shapes introduce a novel fracture region, shown in Figure 3(c), that is not present in the corresponding complete shape, by learning to infer a restoration shape code from an input fractured shape, as shown on the left of Figure 1. Though a small body of prior work exists to generate restorative parts for fractured objects, these approaches depend on prior knowledge of shape structure, e.g., a precise pre‐existing 3D replica [LBB19] that may not always exist, or require that the object be symmetrical and the damage occur asymmetrically [SLL∗08; GSP∗14; PSA∗17]. The work of Hermoza and Sipiran [HS18] is the only known approach to generate repair parts for damaged shapes without structural constraints.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

MendNet: Restoration of Fractured Shapes Using Learned Occupancy Functions

Lamb

Banerjee

2022

Computer Graphics Forum

View full text Add to dashboard Cite

We provide a novel approach to perform fully automated generation of restorations for fractured shapes using learned implicit shape representations in the form of occupancy functions. Our approach lays the groundwork to perform automated object repair via additive manufacturing. Existing approaches for restoration of fractured shapes either require prior knowledge of object structure such as symmetries between the restoration and the fractured object, or predict restorations as voxel outputs that are impractical for repair at current resolutions. By leveraging learned occupancy functions for restoration prediction, our approach overcomes the curse of dimensionality with voxel approaches, while providing plausible restorations. Given a fractured shape, we fit a function to occupancy samples from the shape to infer a latent code. We apply a learned transformation to the fractured shape code to predict a corresponding code for restoration generation. To ensure physical validity and well‐constrained shape estimation, we contribute a loss that models feasible occupancy values for fractured shapes, restorations, and complete shapes obtained by joining fractured and restoration shapes. Our work overcomes deficiencies of shape completion approaches adapted for repair, and enables consumer‐driven object repair and cultural heritage object restoration. We share our code and a synthetic dataset of fractured meshes from 8 ShapeNet classes at: https://github.com/Terascale-All-sensing-Research-Studio/MendNet.

show abstract

Section: Limitations and Future Workmentioning

confidence: 99%

Section: Limitations and Future Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

MendNet: Restoration of Fractured Shapes Using Learned Occupancy Functions

Lamb

Banerjee

2022

Computer Graphics Forum

View full text Add to dashboard Cite

show abstract

“…These approaches fail to restore asymmetrical shapes or shapes that have non-symmetric fractures. Lamb et al [24] perform repair without relying on symmetries. However, they require that the complete counterpart be provided as input alongside the fractured shape.…”

Section: Related Workmentioning

confidence: 99%

“…Sub-Lamb. We use the approach of Lamb et al [24] to perform subtraction. Unlike our approach, Lamb et al require a complete counterpart to be provided alongside the fractured shape as input.…”

Section: Comparing Deepmend Restorations To 3d-organ and Baselinesmentioning

confidence: 99%

DeepMend: Learning Occupancy Functions to Represent Shape for Repair

Lamb¹,

Banerjee²,

Banerjee³

2022

Preprint

View full text Add to dashboard Cite

We present DeepMend, a novel approach to reconstruct restorations to fractured shapes using learned occupancy functions. Existing shape repair approaches predict low-resolution voxelized restorations, or require symmetries or access to a pre-existing complete oracle. We represent the occupancy of a fractured shape as the conjunction of the occupancy of an underlying complete shape and the fracture surface, which we model as functions of latent codes using neural networks. Given occupancy samples from an input fractured shape, we estimate latent codes using an inference loss augmented with novel penalty terms that avoid empty or voluminous restorations. We use inferred codes to reconstruct the restoration shape. We show results with simulated fractures on synthetic and real-world scanned objects, and with scanned real fractured mugs. Compared to the existing voxel approach and two baseline methods, our work shows state-of-the-art results in accuracy and avoiding restoration artifacts over non-fracture regions of the fractured shape.

show abstract

DeepMend: Learning Occupancy Functions to Represent Shape for Repair

Lamb

Banerjee

2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Automated reconstruction of smoothly joining 3D printed restorations to fix broken objects

Cited by 12 publications

References 21 publications

MendNet: Restoration of Fractured Shapes Using Learned Occupancy Functions

MendNet: Restoration of Fractured Shapes Using Learned Occupancy Functions

DeepMend: Learning Occupancy Functions to Represent Shape for Repair

DeepMend: Learning Occupancy Functions to Represent Shape for Repair

Contact Info

Product

Resources

About