Assessing the Archaeological and Ecological Legacy of Ancient Southeast Asian Civilisations in Tropical Forests with Multimodal Deep Learning – DEEPFOREST

Despite mounting evidence that ancient human activities have significantly shaped Southeast Asian tropical forests, the extent and nature of this influence remain largely unexplored. Assessing how ancient civilisations impacted these ecosystems is essential to better understand their current dynamics and the long-term consequences of human actions.

Identifying traces of ancient activities and correlating them with current vegetation patterns over large areas is a major challenge due to the density of tropical forests and the subtlety of millennia-old archaeological features. To overcome these obstacles, we propose to augment a large corpus of newly accessible remote sensing data with complementary in-situ measurement and multimodal observations—including LiDAR, orthophotos, satellite imagery, and hyperspectral data. Leveraging these acquisitions, we will develop and train multimodal deep learning models to detect archaeological features and predict key forest functional characteristics and indicators such as biomass, forest types, and vegetation composition.

Our interdisciplinary team, comprising experts in archaeology (EFEO), machine learning (ENPC), forestry (AMAP), and environmental sciences (LSCE), aims to provide new insights into these complex ecosystems and to explore the relationship between ancient occupations and current forest structures, thus informing sustainable ecological practices.