ELDOR: A Dataset and Benchmark for Illegal Gold Mining in the Amazon Rainforest
Quick Answer
ELDOR is a new UAV benchmark dataset for monitoring illegal gold mining in the Amazon, featuring over 2,500 hectares of annotated imagery.
Quick Take
ELDOR is a new UAV benchmark dataset for monitoring illegal gold mining in the Amazon, featuring over 2,500 hectares of annotated imagery. It establishes four benchmark tasks and highlights the challenges of detecting small-scale mining structures, emphasizing the need for advanced multimodal modeling techniques.
Key Points
- ELDOR contains pixel-level semantic labels for mining activities and ecological structures.
- Four benchmark tasks include semantic segmentation and multi-label classification.
- Current models struggle with rare small-scale mining structures and fine-grained classes.
- An interactive explorer tool is provided for domain experts for data exploration.
- The dataset aims to improve monitoring of environmental impacts from illegal mining.
Paper Resources
📖 Reader Mode
~2 min readAuthors:Kangning Cui, Surendra Bohara, Suraj Prasai, Zishan Shao, Wei Tang, Martin Pillaca, Edwin Flores, Zhen Yang, Gregory Larsen, Evan Dethier, David Lutz, Jean-Michel Morel, Miles Silman, Victor Pauca, Fan Yang
Abstract:Illegal gold mining in the Amazon rainforest causes deforestation, water contamination, and long-term ecosystem disruption, yet remains difficult to monitor at fine spatial scales. Satellite imagery supports large-scale observation, but often misses small mining-related structures and subtle land-cover transitions, especially under frequent cloud cover. We introduce ELDOR, a large-scale UAV benchmark for monitoring environmental and landscape disturbance from illegal gold mining in the rainforest. ELDOR contains manually annotated orthomosaic imagery covering over 2,500 hectares, with pixel-level semantic labels for both mining-related activities and surrounding ecological structures. With this unified annotation source, we establish four benchmark tasks: semantic segmentation, segmentation-derived recognition, direct multi-label classification, and class-presence recognition with vision-language models. Across these tasks, we compare generic and remote-sensing-specific segmentation models, vision foundation model-related segmentation methods, direct multi-label classification methods, and vision-language models under a controlled closed-set protocol. Results show that current methods still struggle with rare small-scale mining structures and fine-grained recovery classes, suggesting the need for context-aware and multimodal modeling. To support domain analysis and practical use, we further build an interactive explorer for domain experts that provides a unified interface for data exploration and model inference.
| Comments: | 70 pages, 35 figures, 28 tables |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2605.15397 [cs.CV] |
| (or arXiv:2605.15397v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2605.15397 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Kangning Cui [view email]
[v1]
Thu, 14 May 2026 20:30:25 UTC (20,410 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →LLM-Guided ANN Index Optimization for Human-Object Interaction Retrieval
A phase-aware LLM agent optimizes human-object interaction retrieval, outperforming Optuna TPE by 33.3% and VDTuner by 34.2% on the HICO-DET benchmark. This method enhances throughput by 15.3x over UniIR and demonstrates strong transferability across vector database management systems.