update readme

single_view_recon/README.md

@@ -1,21 +1,51 @@
-## 主要目录
-
-    .
-    ├── model                           #存放网络的模块定义, dataloader等。
-    ├── utils                           #存放一些utility function，如文件操作，网格处理等。
-    ├── evaluate_cascade.py             # code for evaluating cascaded network       
-    ├── evaluate_3Pole.py               # code for evaluating one-stage network 
-    ├── train_stage1.py                 # code for training stage 1 of cascaded framework    
-    ├── train_stage2.py                 # code for training stage 2 of cascaded framework 
-    ├── retrieve_stage1_result.py       # code for retrieving stage-1 results
-    ├── stage2_trainer.py               # callable training module for training 3-category classification 
-    ├── train_3PoleSDF_in_one_stage.py  # code for training one-stage framework
-    ├── finetune_model.py               # code for fine tuning pretrained model
-
-
-
-
-
-
-
-
+# #3PSDF-SVR
+
+This repository contains the source code for applying 3PSDF to 3D reconstruction tasks based on single-view images.
+
+
+## Code
+### Installation
+This repository depends on [Tensorflow](https://pytorch.org/), [NumPy](https://numpy.org/), [Scikit-image](https://scikit-image.org/) and [Horovod](https://github.com/horovod/horovod). The code is tested under the following package versions with CUDA 11.2 and Ubuntu 18.04:
+```
+tensorflow-gpu==2.6.0
+numpy==1.19.5
+scikit-image==0.18.3
+horovod==0.23.0
+```
+
+### Training
+
+* Example command with required parameters to indicate the data folder:
+```
+horovodrun -np 2 python -m src.train
+  --sdf_dir data/sdf-depth7-tfrecord
+  --cam_dir data/cam-tfrecord
+  --img_dir data/img-tfrecord
+  --split_file data/datasplit/train.lst
+``` 
+* Use ```horovodrun -np GPU_NUM``` to indicate the number of GPUs used for distributed training. If you only have one GPU, set ```GPU_NUM``` to 1. You are recommended to use multiple GPUs. 
+* See `python -m src.train --help` for all the detailed training options.
+
+### Testing
+* Example command with required parameters to indicate the data folder and pre-trained model:
+```
+python -m src.test
+  --sdf_dir data/sdf-depth7-tfrecord
+  --cam_dir data/cam-tfrecord
+  --img_dir data/img-tfrecord
+  --split_file data/datasplit/test.lst
+  --load_model_path weights/3psdf_svr_weights
+``` 
+* See `python -m src.test --help` for all the detailed testing options. 
+
+### Data & Download 
+* Train/test data (in TFRecord format): [data.zip](https://drive.google.com/file/d/1sNU1av82qrEMq0mTQVl-e_4eo0mkKXsZ/view?usp=sharing).
+* Pre-trained model: [weights.zip](https://drive.google.com/file/d/1pPboSALmvZJTEjFLgmKBXm32RNVVR1zt/view?usp=sharing).
+* Raw ShapeNet meshes with consistent normals: [shapenet_consistent_normal.zip]()
+* 3PSDF values sampled for ShapeNet shapes: [shapnet_3psdf.zip]()
+* Raw ShapeNet rendered images and camera parameters: [shapenet_renderings.zip]()
+* We provide example scripts to convert the raw data to TFRecord in ```src/utils/shapenet_tfrecord_generator.py```
+
+
+## Contact
+If you have any questions, please email [Cheng Lin](https://clinplayer.github.io/) at [email protected].