5358-UR 3D Object Reconstruction from a Single Image

Reconstructing a complete 3D object from a single image is challenging because the image provides only a partial view of the object’s shape. This project explores a contour-based approach to reconstructing 3D objects from a single RGB image. The proposed pipeline represents image contours using lines and curves and uses these geometric features to predict a point cloud, a set of points describing the object in three-dimensional space. The current prototype focuses on contour encoding and point prediction using a transformer network. The broader architecture proposes selectively applying a generative model to complete hidden regions, with the goal of avoiding unnecessary computation. The project investigates whether contour representations provide sufficient information for shape reconstruction and whether selective completion can reduce computational cost while maintaining reconstruction quality. Future work will evaluate reconstruction accuracy, hidden-region completion, and computational efficiency on unseen objects, and extend the predicted point clouds into surface meshes with color.

Authors

Publication Details

Journal
DigitalCommons - Kennesaw State University (Kennesaw State University)
Published
2026-11-20
Primary Topic
3D Shape Modeling and Analysis
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

5358-UR 3D Object Reconstruction from a Single Image

Acey Dufresne
DigitalCommons - Kennesaw State University (Kennesaw State University)
3D Shape Modeling and Analysis
article

5358-UR 3D Object Reconstruction from a Single Image

Acey Dufresne
article en

Abstract

Reconstructing a complete 3D object from a single image is challenging because the image provides only a partial view of the object’s shape. This project explores a contour-based approach to reconstructing 3D objects from a single RGB image. The proposed pipeline represents image contours using lines and curves and uses these geometric features to predict a point cloud, a set of points describing the object in three-dimensional space. The current prototype focuses on contour encoding and point prediction using a transformer network. The broader architecture proposes selectively applying a generative model to complete hidden regions, with the goal of avoiding unnecessary computation. The project investigates whether contour representations provide sufficient information for shape reconstruction and whether selective completion can reduce computational cost while maintaining reconstruction quality. Future work will evaluate reconstruction accuracy, hidden-region completion, and computational efficiency on unseen objects, and extend the predicted point clouds into surface meshes with color.

DigitalCommons - Kennesaw State University (Kennesaw State University)
3D Shape Modeling and Analysis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon