assignment_2_Machine_learning

Machine learning environment for Siebren, Yitong & Fengyan.

Basic Info

Each run needs to establish objects from more than one million points of point cloud files, and then train and test the objects dataset, which is very time-consuming. Therefore, store the generated 500 objects with normalized features as dataset.txt file and the corresponding ground truth label as label.txt file.

Thus we can directly read these two files for SVM and Random Forest algorithm.

All the .py files are in the src folder.

HOW TO USE

• In folder Dataset, dataset.txt and label.txt files should already exist.

• If not, run src\Pre_main.py to build the dataset.txt and label.txt.

• If files already exist, run src\ML_main.py to perform SVM and Random Forest classification.

• --> Features need to be changed?

• --> Change features in src\Pre_main.py and run it, .txt files will be updated. Then use src\ML_main.py to perform classifications.

Dataset folder:

dataset.txt -- store the 500 objects file(with 6 normalized features)

label.txt -- store the ground truth labels of 500 objects file

Figure folder

Store the pictures/screenshots which may be used in report.

python files for performing algorithms

Relevant python files are entitled with "ML".

ML_dataset.py -- functions to store 500 objects and labels as .txt files.

ML_main.py -- set this as startup project and run.

Tips

Relative path is used thus this project can be cloned and run directly without any modifications.

Before you run this project, you can find the packages needed: requirements.txt.

Use pip install -r requirements.txt to install appropriate versions of all dependent packages if you haven't got them.