Pyramid structure discovered near Caral Peru
omniletters.com
omniletters.com
It was a covered structure. A Huaca. Lima has quite a few in the middle of the city. A stark reminder that we’ve been there ,intermittently, for millennia.
Tom Wessels: Reading the Forested Landscape
[Part 1] https://www.youtube.com/watch?v=zcLQz-oR6sw
I know that 1491 was a pretty good book about this. But it's like 20 years old now. And Lidar seems to have really opened up new insights in the past decade or so.
Cabeza de Vaca spent 1528–1536 wandering through the Southwest, living with multiple indigenous tribes. His experiences ranged from enslavement to becoming a medicine man. His firsthand account, Naufragios, is available, but I highly recommend A Land So Strange by Andrés Reséndez for a more accessible read. De Vaca also had a second adventure in South America, but it’s not as well-documented.
Another great read is River of Darkness by Buddy Levy, which covers Francisco Orellana’s journey down the Amazon. His expedition was roughly contemporary to Cabeza de Vaca’s own jungle survival story—though Orellana was a bit more conquistadorial than De Vaca.
I’d also love to see a proper follow-up to 1491 (1493 doesn’t count!). The closest thing we have might be America Before by Graham Hancock, which incorporates recent LiDAR discoveries—but it leans more into speculation than hard archaeology.
I found Pekka Hämäläinen's Indigenous Continent to be a worthwhile history of North America, although mostly it was focused on events during European exploration of North America with a relatively short initial section about the state of civilization prior to the Spanish arrival.
I love it when a plan comes together.
Some initial thoughts:
- Data Augmentation: Using synthetic or simulated LiDAR data from known structures to improve training. - Few-Shot Learning / Transfer Learning: Training models on better-documented archaeological sites and applying them to new areas.
Would love to hear thoughts from people with experience in remote sensing, computer vision, or archaeology!
The basic answer is that you have to have at least some data to be able to do anything. We found that in the data regime augmentation works a little bit, transfer learning much less so. What really works is simply sitting down and annotating the data, with on-the-ground surveys and follows ups.
Here's also an interesting LiDAR segmentation paper for Maya settlements. Smaller scale, but you can see what people are trying: https://arxiv.org/abs/2208.03163
Here's another curious recent LiDAR paper, no computer vision, but shows the importance of the proper data processsing:
Auld-Thomas L, Canuto MA, Morlet AV, et al. Running out of empty space: environmental lidar and the crowded ancient landscape of Campeche, Mexico. Antiquity. 2024;98(401):1340-1358. doi:10.15184/aqy.2024.148