# OpenAI to Z Challenge

Source: https://www.youtube.com/watch?v=0Ka-Je6mIs8
Recap page: https://rapidrecap.app/video/0Ka-Je6mIs8
Generated: 2025-08-28T20:05:34.194+00:00

---
## Quick Overview

Team Black Bean won the OpenAI to Z Challenge with their 'Archaos' project, a deep learning approach to identify archaeological sites in the Amazon rainforest using satellite imagery, LIDAR, and indigenous knowledge.

**Key Points:**
- Team Black Bean won the OpenAI to Z Challenge for their 'Archaos' project, utilizing deep learning to discover archaeological sites in the Amazon.
- The Archaos system processes LIDAR and spectral datasets to identify features like Amazon Dark Earths, Geoglyphs, and Mounds.
- The model achieved high performance metrics: AUC ~0.95, AP ~0.84, and Accuracy ~0.89.
- The project involved creating an interactive map website to visualize potential sites and their data.
- The team used a GPT-based triage system to analyze and prioritize potential archaeological sites.
- Key data inputs included soil phosphorus, elevation, and river distances, which were used to train prediction and detection models.

![Screenshot at 00:22: The presentation slide introduces 'Archaos: Explore the Hidden World with AI', showcasing a diagram of how the system processes LIDAR and spectral datasets, utilizes domain-specialized AI agents, and employs a natural-language chat interface.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-22.png)

**Context:** The video documents the presentation and outcomes of the OpenAI to Z Challenge, focusing on archaeological site discovery in the Amazon rainforest. The challenge aimed to leverage AI and deep learning to identify hidden human settlements and artifacts, which are often obscured by dense vegetation and difficult terrain. Several teams presented their solutions, highlighting innovative approaches to data processing and analysis.

## Detailed Analysis

Team Black Bean's 'Archaos' project, developed for the OpenAI to Z Challenge, utilizes a deep learning approach to identify archaeological sites in the Amazon rainforest. The system processes massive LIDAR and spectral datasets, combined with open-source data and indigenous histories, to uncover hidden sites. Archaos employs domain-specialized AI agents to analyze terrain, environment, and history, presenting findings through a natural-language chat interface. The model achieved impressive performance metrics, including an AUC of approximately 0.95, an AP of 0.84, and an accuracy of 0.89. The project involved creating a scalable system for processing and analyzing data, including customizing LIDAR processing workflows and generating heatmaps of anomaly candidates. They also developed an interactive map website to visualize identified sites. The team's methodology involved generating negative samples for training and using a GPT-based triage system to evaluate potential sites based on various factors like landscape setting, elevation, and historical insights. The project successfully identified over 100 potential sites, demonstrating the effectiveness of their AI-driven approach in accelerating archaeological discovery.

### Project Overview

- Archaos - Explore the Hidden World with AI
- Deep learning approach for Amazon archaeological site discovery
- Utilizes satellite imagery, LIDAR, open-source data, and indigenous histories

### Methodology

- Processing and analyzing LIDAR and spectral datasets
- Domain-specialized AI agents for terrain, environment, and history analysis
- Natural-language chat interface for user interaction
- GPT-based triage for site prioritization

### Key Features & Performance

- Identified various features like Amazon Dark Earths, Geoglyphs, Mounds
- Achieved AUC ~0.95, AP ~0.84, Accuracy ~0.89
- Tessellated Acre into 3x3 km tiles, aggregated to 20 km super-tiles
- Noise reduction and feature visibility enhancement in processed data

### Results & Findings

- Discovered over 100 potential archaeological sites
- Developed an interactive map website for site visualization
- Demonstrated the scalability of the deep learning approach for large-scale analysis

### Team Introduction

- Team Black Bean (Yunxuan Tian, Yao Zhao, Yingjie Zhang)
- Discussion of team's motivations and background in AI and data science
- Explanation of the name 'Black Bean' in honor of a deceased family member

![Screenshot at 00:01: Text overlay stating that the Amazon spans 6 million square kilometers across 9 countries and contains traces of ancient civilizations.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-01.png)
![Screenshot at 00:10: Text overlay announcing the launch of the 'OpenAI to Z Challenge'.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-10.png)
![Screenshot at 00:13: Text overlay detailing the challenge's goal: calling on developers to help archaeologists uncover hidden sites using satellite imagery, LIDAR, open-source data, and indigenous histories.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-13.png)
![Screenshot at 00:22: Presentation slide introducing 'Archaos: Explore the Hidden World with AI', featuring a diagram of the system's components and data processing pipeline.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-22.png)
![Screenshot at 00:27: Screenshot of the Archaos interface, showing a world map with highlighted potential archaeological sites and an AI chat interface.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-27.png)
![Screenshot at 00:31: Slide titled 'WHAT are we looking for?' listing types of archaeological features in Amazonia, such as Amazon Dark Earths, Geoglyphs, and Mounds, with a visual example of a circular earthwork.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-31.png)
![Screenshot at 00:37: Title slide for the project: 'Amazon Archeological Sites Discovery - a Deep Learning Approach', crediting Team Black Bean.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-37.png)
![Screenshot at 00:43: Visual representation of processed data, showing different transformations like 'Raw', 'Smoothed', 'Local StdDev', 'Laplacian', and 'Slope', derived from LIDAR data.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-43.png)
![Screenshot at 00:46: Slide on 'Metrics & scoring' displaying model performance metrics \(AUC, AP, Accuracy\) and a candidate-candidate heat-map.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-46.png)
![Screenshot at 00:55: Visual explanation of why soil phosphorus, elevation, and river distances are important inputs for the prediction and detection models, with illustrative images for each factor.](https://ss.rapidrecap.app/screens/0Ka-Je6mIs8/00-00-55.png)