Skip to content

The project is based on spherical harmonics and fully connected network to classify 2 classes of models in ModelNet40, since the overall accuracy of ModelNet40 is really low.

Notifications You must be signed in to change notification settings

Python-Consultants/3d-model-classify-retrival

Repository files navigation

3d-model-classify-retrival

The project is only for discovering and can't really be used in real-world project

安装的python包见requirement.txt

##主目录
*requirement.txt 程序需要的依赖包可以通过pip install -r requirements.txt安装
*ModelNet2数据集文件夹(来自ModelNet40,选取两个)
*showPlot.py 展示图片的函数
*ModelGeneration.py 生成神经网络模型的文件
*model.h5 生成的模型的保存文件
*model.json 生成的模型的权重参数
*util.py 距离计算的公式的文件
*classifier.py 生成模型的分类和计算最相似的模型
*FeaGeneration.py 生成特征矩阵,被用于ModelGeneration和classifier
*Sph_harm_for2.ipynb 探索的jupyter文件

##运行步骤和命令行命令:
*首先安装需要的依赖
*首先通过cd进入主目录
*运行python ModelGeneration.py 得到model.h5和model.json,和模型输出效果分析
*运行python classifier.py ModelNet2/glass_box/test/glass_box_0246.off 即可对ModelNet2/glass_box/test/glass_box_0246.off这个文件输出原图和五个最相似的图片

About

The project is based on spherical harmonics and fully connected network to classify 2 classes of models in ModelNet40, since the overall accuracy of ModelNet40 is really low.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published