Deep Learning for Geospatial Data:
A Hands-On Workshop
(Looking for the “sharing” workshop? Click here!)
Important!!
You will need a recently made, mid-level “gamer” computer to use Engine and do the exercises in this workshop.
Hardware Requirements
- Processor: 64-bit processor, x86-64, 2 GHz, 2 available processor cores.
- Graphics processor: a GPU that supports Nvidia CUDA 11.1.
- Computer memory: 16 GB RAM.
- Disk storage: 500 GB of space on an SSD
Software Requirements
- Windows 10 or higher, 64-bit version
- The NVIDIA CUDA Toolkit, version 11.1 (only this version)
Why Use Deep Learning?
An example application: Identifying cars in aerial imagery:
Outline
Introduction to Deep Learning
- What is deep learning?
- A typical workflow using deep learning technology to identify objects.
- Issues to be aware of when using deep learning technology.
- Options for deep learning tools.
Applying Deep Learning to Geospatial Data
- Example 1: Using deep learning to identify objects in aerial (orthomosaic) imagery
- Workflow
- Tips
- Hands-on training a model
- Hands-on using a model
- Example 2: Using deep learning to identify objects in panoramic (scene view) imagery
- Workflow
- Tips
- Hands-on training a model
- Hands-on using a model
- Example 3: Using deep learning to classify point cloud data
- Workflow
- Tips
- Hands-on training a model
- Hands-on using a model
Register
Note: We will not accept registrations using Gmail, etc. We will only accept registration from people using their official email.