ArcGIS : Learn Deep Learning in ArcGIS to advance GIS skills

Learn the basics of deep learning with neural networks for image classification and object detection. See how gradient descent and backpropagation optimize model parameters through iterative learning.

Integrate deep learning with ArcGIS Pro to perform pixel and object classification and detection using CNNs, with TensorFlow, PyTorch, and Keras, training data preparation, and inference via model definition files.

Master the end-to-end deep learning workflow in ArcGIS Pro, from creating training samples and image chips to configuring a Python environment and applying inferencing for object detection.

Learn to obtain remote sensing imagery from Open Aerial Map for deep learning object detection in ArcGIS Pro, using a 9 cm UAV image of palm trees on Tonga Island.

Create and export training samples in ArcGIS Pro to train CNNs, using labeled images, polygons, points, or rasters, with PyTorch training and model definition files (.emd).

Open ArcGIS Pro, load a color TIFF image as a true color composite to identify palm trees, then create bookmarks for training samples for a deep learning model.

Create a palm-class training schema and collect palm tree samples in ArcGIS Pro 2.5 to train a deep learning model, then export JPEG data with Pascal VOC metadata.

Clone and activate the default Conda environment to enable deep learning tools in ArcGIS Pro, following copy-paste Python commands and verifying package installations via ArcGIS Pro help.

Create deep learning model definitions in ArcGIS Pro with the tab training tool, yielding an emd json file or json string. Model definitions inform inference tools about inputs and outputs.

Train a cnn deep learning model in ArcGIS Pro using image chips and the single shot detector, with 20 epochs, to produce a palm trees model; learn parameter tuning.

Apply a trained deep learning model in ArcGIS Pro to satellite images for object detection or classification, producing a feature class with bounding boxes, polygons, or centers.

Learn to detect palm trees with a CNN-based deep learning model in ArcGIS Pro, adjusting padding, threshold, max overlap, and non-maximum suppression on image subsets.

Convert polygon palm tree polygons to points, buffer each point by three meters, and color buffers red to visually enhance detected objects on the map.

Explore geospatial predictive modeling and deep learning in GIS, using regression, kriging, and AI to analyze big data, forecast events, and enhance satellite imagery.

Create an independent training set for buildings in the UAV image and detect them with deep learning in ArcGIS Pro, following the palm trees workflow.

Explore how deep learning enhances GIS with ArcGIS Pro's user-friendly tools for supervised learning, classification, and object detection, without programming skills, while noting limited parameter control and a black-box model.

Explore further learning in remote sensing, data science, and machine learning via the Udemy instructor page and the Jio World YouTube channel, and follow social media for updates.