File size: 2,525 Bytes
d2eed81 924fcea d2eed81 924fcea | 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 | ---
library_name: pytorch
base_model: facebook/VGGT-1B
datasets:
- czZhang2026/ViewRecDB-100K
tags:
- computer-vision
- viewpoint-recommendation
- ai-photography
---
<div align="center">
<h1>ViewRecNet</h1>
<p><strong>ViewRec3D: Learning to Recommend 3D Viewpoints for AI Photography</strong></p>
<p>
<a href="https://github.com/anonymous1-submission/ViewRec3D/blob/main/ViewRec3D.pdf"><img src="https://img.shields.io/badge/Paper-ViewRec3D-blue" alt="Paper"></a>
<a href="https://github.com/anonymous1-submission/ViewRec3D"><img src="https://img.shields.io/badge/GitHub-Code-24292f" alt="Code"></a>
<a href="https://huggingface.co/datasets/czZhang2026/ViewRecDB-100K"><img src="https://img.shields.io/badge/Dataset-ViewRecDB--100K-orange" alt="Dataset"></a>
</p>
<p>Chengzhi Zhang<sup>1,*</sup>, Yunzhong Hou<sup>1,*,†</sup>, Zizheng Sun<sup>1</sup>, Ming-Hsuan Yang<sup>2</sup>, Chi Harold Liu<sup>1</sup></p>
<p><sup>1</sup> Beijing Institute of Technology <sup>2</sup> University of California, Merced</p>
<p><sup>*</sup> Equal contribution. <sup>†</sup> Corresponding author.</p>
</div>
## Overview
Given a single photograph, ViewRecNet recommends how to move and rotate the camera, adjust the field of view, and choose portrait or landscape orientation to approach an expert composition. The model uses a VGGT backbone and learns from paired views in ViewRecDB-100K with dense camera-ray supervision.
## Released checkpoint
`model.safetensors` (4.51 GB) contains the weights from training step 15,625, including the backbone, camera head, orientation head, and ray head. The evaluation code constructs the network and loads these weights directly.
## Quick start
From the ViewRec3D code repository, install the project requirements and download the checkpoint:
```bash
python -m pip install -r requirements.txt
python -m pip install -U huggingface_hub
hf download czZhang2026/ViewRecNet model.safetensors --local-dir checkpoints/ViewRecNet
```
To evaluate on a CUDA GPU, prepare the test split as described in the [dataset card](https://huggingface.co/datasets/czZhang2026/ViewRecDB-100K). Replace the test directory below with its local path. The evaluator reads the checkpoint from `--output_dir` and writes evaluation metrics there.
```bash
python -m src.evaluate --val_dir /path/to/ViewRecDB-100K/test --output_dir checkpoints/ViewRecNet --use_rays True
```
## Acknowledgements
This work was supported by the National Natural Science Foundation of China (No. U25A20434).
|