Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
张佳预 192d7a08d5 | 1 year ago | |
---|---|---|
images | 1 year ago | |
.gitignore | 1 year ago | |
LICENSE | 1 year ago | |
README.md | 1 year ago | |
README_EN.md | 1 year ago | |
cp_material.sh | 1 year ago | |
dataset.py | 1 year ago | |
dpn.py | 1 year ago | |
ednet.py | 1 year ago | |
stn.py | 1 year ago | |
test.py | 1 year ago | |
train.py | 1 year ago | |
utils.py | 1 year ago | |
visualizer.py | 1 year ago |
简体中文|English
Tiancheng Zhi, Bernardo R. Pires, Martial Hebert, Srinivasa G. Narasimhan
这是CVPR代码的改进和简化版本。与原始CVPR版本相比,这段代码有更好的性能,主要的改进有:
STN没有白平衡
在STN中使用普通卷积而不是对称卷积
随机翻转STN的输入和输出
自动调整学习率
超参数有调整
要与原始CVPR结果进行比较,请参阅项目页面(数据集的第一个下载链接)。
TITAN Xp GPU * 2
Ubuntu 16.04
Python 3
PyTorch 1.0
OpenCV
Visdom (for visualization)
下载rgbnir_stereo,将data
和lists
移动到DMC
文件夹中。
下载precomputed_material,并把它放在DMC
文件夹下。
然后运行:
sh cp_material.sh precomputed_material data
更多关于匹兹堡立体数据集的信息和下载链接,请参阅项目页面。
CUDA_VISIBLE_DEVICES=1,0 python3 train.py
CUDA_VISIBLE_DEVICES=1,0 python3 test.py --ckpt-path ckpt/47.pth
下载预训练模型 pretrained.pth
性能 (RMSE,值越小越好):
模型 | 普通场景 | 灯光 | 玻璃 | 光滑表面 | 植物 | 皮肤 | 衣服 | 包 | 平均 |
---|---|---|---|---|---|---|---|---|---|
CVPR'18 | 0.53 | 0.69 | 0.65 | 0.70 | 0.72 | 1.15 | 1.15 | 0.80 | 0.80 |
Pretrained | 0.47 | 0.56 | 0.56 | 0.61 | 0.72 | 0.93 | 0.91 | 0.86 | 0.70 |
CVPR2018无监督RGB-NIR跨光谱立体匹配
Python Markdown Shell
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》