Pith. sign in

REVIEW 2 cited by

DreamCar: Leveraging Car-specific Prior for in-the-wild 3D Car Reconstruction

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2407.16988 v2 pith:NBTQUUT5 submitted 2024-07-24 cs.CV

DreamCar: Leveraging Car-specific Prior for in-the-wild 3D Car Reconstruction

classification cs.CV
keywords carsdatasetsgenerativehigh-qualityimagesmethodavailablecar-specific
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Self-driving industries usually employ professional artists to build exquisite 3D cars. However, it is expensive to craft large-scale digital assets. Since there are already numerous datasets available that contain a vast number of images of cars, we focus on reconstructing high-quality 3D car models from these datasets. However, these datasets only contain one side of cars in the forward-moving scene. We try to use the existing generative models to provide more supervision information, but they struggle to generalize well in cars since they are trained on synthetic datasets not car-specific. In addition, The reconstructed 3D car texture misaligns due to a large error in camera pose estimation when dealing with in-the-wild images. These restrictions make it challenging for previous methods to reconstruct complete 3D cars. To address these problems, we propose a novel method, named DreamCar, which can reconstruct high-quality 3D cars given a few images even a single image. To generalize the generative model, we collect a car dataset, named Car360, with over 5,600 vehicles. With this dataset, we make the generative model more robust to cars. We use this generative prior specific to the car to guide its reconstruction via Score Distillation Sampling. To further complement the supervision information, we utilize the geometric and appearance symmetry of cars. Finally, we propose a pose optimization method that rectifies poses to tackle texture misalignment. Extensive experiments demonstrate that our method significantly outperforms existing methods in reconstructing high-quality 3D cars. \href{https://xiaobiaodu.github.io/dreamcar-project/}{Our code is available.}

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Monte Carlo Energy Aggregation for Mobile 3D Gaussian Splatting

    cs.CV 2026-06 unverdicted novelty 6.0

    Flux-GS is a mobile-optimized 3D Gaussian Splatting method that compresses specular energy via Monte Carlo aggregation, recovers details with attribute-conditioned SH offsets, and uses multi-view guidance for densific...

  2. SynthDrive: Scalable Real2Sim2Real Sensor Simulation Pipeline for High-Fidelity Asset Generation and Driving Data Synthesis

    cs.CV 2025-09 conditional novelty 5.0

    SynthDrive automatically mines images of rare objects, reconstructs them as 3D assets from a single view, and synthesizes driving footage that modestly improves detection of those objects.