Created by Hang Zhang from Northwestern Polytechnical University.

ASIN is a novel multi-task network based on point cloud data for machining feature recognition. ASIN realizes machining feature segmentation, machining feature identification, and bottom face identification simultaneously. The final machining feature recognition results are obtained by combining the results of the above three tasks.

In this repository, we release source codes and for ASIN. It's also worth noting that, because to the necessity for secrecy, we only release a portion of the part point cloud data that contains holes, pockets, and slots.

(1) cuda 11.0.3
(2) python 3.8.8
(3) tensorflow 2.4.0
(4) h5py 2.10.0
(5) scikit-learn 0.24.1
(6) numpy 1.19.2
(7) matplotlib 3.3.4

The code is tested on Intel Core i9-10980XE CPU, 128GB memory, and NVIDIA GeForce RTX 3090 GPU.

(1) Get the ASIN source code by cloning the repository: https://github.com/HARRIXJANG/ASIN-master.git.
(2) Create the folder named logdir in the root directory.
(3) Download related point cloud datasets (traindata.h5, validationdata.h5).
(4) Put the datasets in the folder data.
(5) Run python train.py to train the network.

(1) Get the ASIN source code by cloning the repository: https://github.com/HARRIXJANG/ASIN-master.git.
(2) Download related point cloud datasets (testdata.h5).
(3) Put the datasets in the folder data.
(4) Download related pre-trained ASIN model (.h5). The pre-trained ASIN model was trained on the publicly available dataset.
(5) Put pre-trained ASIN model in the folder models.
(6) Run python test.py to test the network.

(1) Get the ASIN source code by cloning the repository: https://github.com/HARRIXJANG/ASIN-master.git.
(2) Download related pre-trained ASIN model (.h5). The pre-trained ASIN model was trained on the publicly available dataset.
(3) Put the pre-trained ASIN model in the folder models.
(4) Run python predict.py to predict a part. The result is a text file (.txt). In this text file, the first line is "start", the second line is the name of the part, the third line is the tag number of each face in the part (generated by catia), the fourth line is the category (hole, pocket) corresponding to each face, the fifth line is the identification of the bottom surface (0 represents a non-bottom face, 1 represents a bottom face), the sixth line is the clustering results (each set of faces represents a machining feature), and the eighth line is "end".

(1) Run python draw_results.py to visualize the result.
(2) Run python draw_original_point_clouds.py to visualize the part point cloud in the dataset.

If you use this code please cite:

@inproceedings{zhang2022asin,  
      title={Machining feature recognition based on a novel multi-task deep learning network},  
      author={Hang Zhang, Shusheng Zhang, Yajun Zhang, Jiachen Liang, and Zhen Wang},  
      booktitle={Robotics and Computer-Integrated Manufacturing},  
      year={2022}  
    }

If you have any questions about the code, please feel free to contact me ([email protected]).