Introduction of Custom Deep Learning Point Cloud Classification
The custom deep learning point cloud classification feature allows you to classify uncategorized point clouds using deep learning point cloud classification models. It also provides the capability to train your own deep-learning point cloud classification models. With this feature, you can manually edit a small amount of data within the same batch and use the edited results to train the model, which can then automatically process a large amount of data.
The "Deep Learning Classification" feature already enables the classification of point clouds into specific categories such as ground points, buildings, vegetation, etc. The "Custom Deep Learning Classification" opens up numerous parameter settings and options, allowing you to classify point clouds into any desired categories. It includes both fully supervised and weakly supervised methods, where fully supervised requires complete annotation of the point cloud data for training, while weakly supervised only requires a small amount of annotated point cloud data, thus significantly improving production efficiency.
This feature supports two main workflows:
(1)Training Sample Creation -> Model Training
(2)Classification Using an Existing Model