Fine-tune the model

You can fine-tune the Building Point Classification model to suit your specific 3D application. Fine-tuning a model requires less training data, computational resources, and time compared to training a new model.

Fine-tuning is recommended if the available ArcGIS pretrained deep learning models don't yield satisfactory results. This might occur when your input dataset's properties differ significantly from the pretrained model's training data in terms of density, sensor or source type, geometric properties of the objects of interest, or major changes in terrain type.

You can use the Prepare Point Cloud Training Data tool to prepare training data. Next, you can fine-tune this model on your data using the Train Point Cloud Classification Model tool in ArcGIS Pro. Follow the steps below to fine-tune the model.

Prepare training data

This model is trained on a classified point cloud, in which no attributes are considered during training. The class codes of interest for the pretrained model are 0 and 6, with 0 representing background points and 6 representing building points. Use the same class structure to prepare the fine-tuning training data using the Prepare Point Cloud Training Data tool.

1. Browse to Tools under the Analysis tab.

   

2. Click the Toolboxes tab in the Geoprocessing pane, select 3D Analyst Tools and browse to the Prepare Point Cloud Training Data tool in the Point Cloud toolset.

   

3. Set the required variables under the Parameters tab as follows:
   1. For Input Point Cloud, select the training LAS dataset.
   2. For Validation Point Cloud, select the validation LAS dataset.
   3. For Output Training Data, set the location and name of the output training data (.pctd).
   4. Set Block Size to 100 meters, same as what is used for the pretrained model.
   5. Set Block Point Limit to 30000, same as what is used for the pretrained model.

      

4. Click Run.

   Once processing is complete, the exported training data is saved in the specified directory.

Fine-tune the Building Point Classification model

Complete the following steps to fine-tune the model:

1. Browse to Tools under the Analysis tab.

   

2. In the Geoprocessing pane, click the Toolboxes tab and expand 3D Analyst Tools. Select the Train Point Cloud Classification Model tool under the Point Cloud toolset.

   

3. Set the required variables under the Parameters tab as follows:
   1. For Input Training Data, set the location of the training data (.pctd).
   2. For Pre-trained Model, provide the path to the pretrained model.
   3. For Output Model Location, set the folder path that will store the new directory containing the deep learning model.
   4. For Output Model Name, provide the name of the output model.
   5. Set Batch Size to 2 Increase or decrease this number according to your GPU capacity.

      Batch size must be a square number.

      

4. Set the variables under the Environments tab.
   1. For Processor Type, select CPU or GPU as needed.

      If GPU is available, it is recommended that you select GPU and set GPU ID to the GPU to be used.

      

5. Click Run.

   You can now use this model to run inferencing against your point cloud dataset.