The identification and characterization of various rock types is one of the
fundamental activities for geology and related areas such as mining, petroleum,
environment, industry and construction. Traditionally, a human specialist is
responsible for analyzing and explaining details about the type, composition,
texture, shape and other properties using rock samples collected in-situ or
prepared in a laboratory. The results become subjective based on experience, in
addition to consuming a large investment of time and effort. The present
proposal uses artificial intelligence techniques combining computer vision and
natural language processing to generate a textual and verbal description from a
thin section image of rock. We build a dataset of images and their respective
textual descriptions for the training of a model that associates the relevant
features of the image extracted by EfficientNetB7 with the textual description
generated by a Transformer network, reaching an accuracy value of 0.892 and a
BLEU value of 0.71. This model can be a useful resource for research,
professional and academic work, so it has been deployed through a Web
application for public use.

利用人工智能技术结合计算机视觉和自然语言处理，从岩石的薄片图像中生成文本和口头描述，通过 EfficientNetB7 提取的图像相关特征与 Transformer 网络生成的文本描述相关联的模型训练，达到了 0.892 的准确度和 0.71 的 BLEU 值。该模型可用作研究、专业和学术工作的有用资源，并已通过网络应用程序供公众使用。

岩石薄片的自动描述：一个 Web 应用

Descripción automática de secciones delgadas de rocas: una  aplicación Web

Modeling distributions on Riemannian manifolds is a crucial component in
understanding non-Euclidean data that arises, e.g., in physics and geology. The
budding approaches in this space are limited by representational and
computational tradeoffs. We propose and study a class of flows that uses convex
potentials from Riemannian optimal transport. These are universal and can model
distributions on any compact Riemannian manifold without requiring domain
knowledge of the manifold to be integrated into the architecture. We
demonstrate that these flows can model standard distributions on spheres, and
tori, on synthetic and geological data. Our source code is freely available
online at this http URL

我们提出并研究了一类使用来自黎曼最优传输的凸势的流，可以对任何紧黎曼流形上的分布进行建模，用于理解物理和地质等非欧几里德数据的分布模拟。