Skip to content

Latest commit

 

History

History
 
 

tensorflow

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
cfgs
cfgs
 
 
datasets
datasets
 
 
function
function
 
 
models
models
 
 
ops
ops
 
 
utils
utils
 
 
README.md
README.md
 
 
init.sh
init.sh
 
 

README.md

Installation

Preparation

Requirements

  • Ubuntu 16.04
  • Anaconda with python=3.6
  • tensorFlow=1.12
  • cuda=9.0
  • cudnn>=7.4
  • others: pip install termcolor opencv-python toposort h5py easydict

Compile custom operators

sh init.sh

Datasets

Shape Classification on ModelNet40

You can download ModelNet40 for here (1.6 GB). Unzip and move (or link) it to data/ModelNet40/modelnet40_normal_resampled.

Part Segmentation on PartNet

You can download PartNet dataset from the ShapeNet official webpage (8.0 GB). Unzip and move (or link) it to data/PartNet/sem_seg_h5.

Scene Segmentation on S3DIS

You can download the S3DIS dataset from here (4.8 GB). You only need to download the file named Stanford3dDataset_v1.2.zip, unzip and move (or link) it to data/S3DIS/Stanford3dDataset_v1.2.

The file structure should look like:

<tf-code-root>
├── cfgs
│   ├── modelnet
│   ├── partnet
│   └── s3dis
├── data
│   ├── ModelNet40
│   │   └── modelnet40_normal_resampled
│   │       ├── modelnet10_shape_names.txt
│   │       ├── modelnet10_test.txt
│   │       ├── modelnet10_train.txt
│   │       ├── modelnet40_shape_names.txt
│   │       ├── modelnet40_test.txt
│   │       ├── modelnet40_train.txt
│   │       ├── airplane
│   │       ├── bathtub
│   │       └── ...
│   ├── PartNet
│   │   └── sem_seg_h5
│   │       ├── Bag-1
│   │       ├── Bed-1
│   │       ├── Bed-2
│   │       ├── Bed-3
│   │       ├── Bottle-1
│   │       ├── Bottle-3
│   │       └── ...
│   └── S3DIS
│       └── Stanford3dDataset_v1.2
│           ├── Area_1
│           ├── Area_2
│           ├── Area_3
│           ├── Area_4
│           ├── Area_5
│           └── Area_6
├── init.sh
├── datasets
├── function
├── models
├── ops
└── utils

Usage

Training

ModelNet

python function/train_evaluate_modelnet.py --cfg <config file>  \
    [--gpus <list of gpus>] [--log_dir <log directory>]
  • <config file> is the yaml file that determines most experiment settings. Most config file are in the cfgs directory.
  • <list of gpus> means the indexes of gpus you want to use for training, like 0, 0 1, 0 1 2 3.
  • <log directory> is the directory that the log file, checkpoints will be saved, default is log.

PartNet

python function/train_evaluate_partnet.py --cfg <config file>  \
    [--gpus <list of gpus>] [--log_dir <log directory>]

S3DIS

python function/train_evaluate_s3dis.py --cfg <config file>  \
    [--gpus <list of gpus>] [--log_dir <log directory>]

Evaluating

ModelNet40

python function/train_evaluate_modelnet.py --cfg <config file> --load_path <checkpoint> \
    [--gpu <gpu>] [--log_dir <log directory>]
  • <config file> is the yaml file that determines most experiment settings. Most config file are in the cfgs directory.
  • <checkpoint> is the model checkpoint used for evaluating.
  • <gpu> means which gpu you want to use for evaluating, note that we only use one gpu for evaluating.
  • <log directory> is the directory that the log file, checkpoints will be saved, default is log_eval.

PartNet

python function/evaluate_partnet.py --cfg <config file> --load_path <checkpoint> \
    [--gpu <gpu>] [--log_dir <log directory>]

S3DIS

python function/evaluate_s3dis.py --cfg <config file> --load_path <checkpoint> \
    [--gpu <gpu>] [--log_dir <log directory>]

Models

Method ModelNet40 S3DIS PartNet (val/test)
Point-wise MLP 92.8 66.2 48.1/51.2
Pseudo Grid 93.0 65.9 50.8/53.0
Adapt Weights 93.0 66.5 50.1/53.5
PosPool 92.9 66.5 50.0/53.4
PosPool* 93.2 66.7 50.6/53.8