Hunt Dinosaurs with Intel® AI

Get the Latest on All Things CODE



June 10, 2022 – In honor of the final Jurassic Park movie, Jurassic World Dominion, let's talk about integrating ancient and modern worlds—hunting dinosaur bones and working with AI.

In 1996, I made the pilgrimage to Dinosaur National Monument (on the Jensen, Utah side) to see their dinosaur bones. Some of the bones I saw were those of a Camarasaurus, a large, long-necked dinosaur that ate plants. Figure 1 shows a Camarasaurus.

Figure 1. Camarasaurus skull.
(Picture courtesy of Bob Chesebrough, technical evangelist for AI for oneAPI, Intel Corporation.)

Computer Vision Models Find Jurassic-Period Bones

To aid with finding dinosaur bones, other data must be used to infer their location. Bone hunters need to look for geological depositional environments that would promote the creation of fossils. In the western United States, dinosaur bones have been found in the Morrison Formation, a banding of rock layers, limestone, and clay formed in the Upper Jurassic period. Visually the layers look like cake sitting atop melted ice cream, shown in Figure 2.

Figure 2. Erosion of sandstone layers on top of clay soils.
(Picture courtesy of Bob Chesebrough.)

Scientists have estimated that there are still thousands of undiscovered dinosaur bones out there.

Wondering where they might be? What if you could create your own dinosaur-bone bed map or a “likelihood map” of where to find them? A dinosaur bone likelihood map, based on the Dinosaur National Monument, was created using computer vision models (shown in Figure 3). To help narrow a bone hunter’s search, the areas in green show where dinosaur bones are likely to be found.

Figure 3. Sample dinosaur bone likelihood map

This likelihood map was created by using data from aerial photographs of the Morrison Formation and the GPS coordinates for small bone fragments discovered there. Images were inputted into a convolutional neural network (CNN), a deep-learning network to train computer vision models to classify the images for their likelihood of having bones.

Before you strike out on your own dinosaur bone hunt, think about creating a likelihood map to guide your efforts.

Think You’ve Found a Dinosaur Bone?

Here's what to do next:

  • Do not dig it up. Dinosaur bones are often protected wherever they’re found, such as on U.S. federal lands.
  • Do take pictures, note the location, and contact a paleontologist through your local natural history museum.

What Is Computer Vision?

This is a field within AI that trains mathematical models to extract information from visual data (such as images or videos) and then provides an interpretation that's similar to how the human visual cortex operates. This field is exciting because it can automate visual tasks that humans may have difficulties with or not be able to perform under certain conditions.
Learn more about computer vision > 

Explore Intel® AI Tools & Training Opportunities


For the software development tools, languages, and frameworks to help you solve your AI needs, see AI and Machine Learning.


July 12, 2022 — Register for the virtual (and free) 2022 oneAPI DevSummit for AI to accelerate AI pipelines and enhance productivity, including how to create a dinosaur bone likelihood map. This day of discovery with renowned industry experts will demystify the latest AI technologies, tools, trends, and techniques.

In addition to the workshop Hunting Dinosaurs with Intel AI Software, conducted by Intel's resident dinosaur expert Bob Chesebrough, several industry leaders from Accenture, RedHat*,*, and others in the AI ecosystem talk about their work with enabling a wide range of AI applications and discuss the developer benefits of working with AI software from Intel.


You May Also Like


Behind the Scenes: Arming Researchers with Visualization Know-How 

SciVis Unveils a Billion Cells, COVID-19, and Invisible Monsters 


Accelerate AI Deep Learning with oneAPI Deep Neural Network Library 

Intel Employs Machine Learning to Creating Computer Graphics 

A Scale-Out Training Solution for Deep-Learning Recommender Systems 


The University of Chicago Uses Scientific Visualization (SciVis) for Healthcare Research That Includes COVID-19


Performance & Scalability of Deep-Learning Inference Based on Convolutional Neural Networks (CNN) 

Optimize Data-Science & Machine-Learning Pipelines 

Intel® oneAPI AI Analytics Toolkit
Accelerate end-to-end machine learning and data science pipelines with optimized deep learning frameworks and high-performing Python* libraries.

Get It Now

See All Tools