pqh22/ProxyTransformation
[CVPR2025] ProxyTransformation : Preshaping Point Cloud Manifold With Proxy Attention For 3D Visual Grounding
This project helps improve the accuracy of pinpointing objects in 3D scenes created from multiple camera views, especially when the initial 3D scan is noisy or lacks detail. It takes in rough 3D point cloud data, along with descriptive text and original multi-view images, to produce a more precise 3D representation with objects accurately located. This would be used by robotics engineers, autonomous vehicle developers, or augmented reality creators who need highly accurate 3D object localization from sensor data.
No commits in the last 6 months.
Use this if you need to enhance the geometric and semantic detail of 3D point clouds for precise object localization using multi-modal information like text descriptions and images.
Not ideal if you are only working with a single modality (e.g., just point cloud data) or if you require a solution that doesn't involve complex multi-modal input processing.
Stars
49
Forks
1
Language
Python
License
MIT
Category
Last pushed
Sep 02, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/pqh22/ProxyTransformation"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
drprojects/superpoint_transformer
Official PyTorch implementation of Superpoint Transformer introduced in [ICCV'23] "Efficient 3D...
yuxumin/PoinTr
[ICCV 2021 Oral] PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers
charlesq34/frustum-pointnets
Frustum PointNets for 3D Object Detection from RGB-D Data
drprojects/DeepViewAgg
[CVPR'22 Best Paper Finalist] Official PyTorch implementation of the method presented in...
facebookresearch/votenet
Deep Hough Voting for 3D Object Detection in Point Clouds