Lightning Talk: Dinosaur Bone Hunt
summary
Builds a probabilistic heat map from tiled aerial images (224x224–512x512) using ResNet18 trained in PyTorch and accelerated with the Intel AI Analytics Toolkit (IPEX) running on CPUs and XPUs. Produces georeferenced likelihood scores, put on a phone map, and tested by visiting high‑score pixels in the Morrison Formation and Dinosaur National Monument where the team finds multiple new bones including ribs and a tibia. Code and training data pipeline are on GitHub and the method proves effective in field trials.
AI-written summary · about
related talks
39mLightning Talks
22mIdentifying Potential Historic Sheep Dip Locations in Southland usi
45mFrom Radar Echoes To Real-Time Recognition: Deep Learning on Another Planet an Open Source Workflow
31mFrom Radar Echoes To Real-Time Recognition: Deep Learning on Another Planet -An Open Source Workflow
21mNeedle in the Haystack: Applied ML for Inspections of Power Infrastructure
23mDeploying machine learning models for forensic anthropology
43mLightning Talks
54mHow to handle the luxury of having too much training data