heig-vd-geo/projet-piscine

该项目是一个学生项目，旨在通过深度学习技术对游泳池进行语义分割。项目涵盖了从地理空间数据准备到最终在HuggingFace排行榜上评估的整个深度学习项目流程。学生将获得包含所有泳池多边形的shapefile、州的正射影像、用于训练的图像集以及无标签的测试图像集。项目要求学生提交科学小论文/工程师报告、功能模型代码以及在HuggingFace排行榜上提交的预测结果。

This project is a student initiative aimed at semantic segmentation of swimming pools using deep learning techniques. It covers the entire pipeline of a deep learning project applied to geomatics, from the preparation of geospatial data to the final evaluation on a HuggingFace leaderboard. Students are provided with a shapefile containing all swimming pool polygons, orthophotos of the canton, a set of training images, and a set of unlabeled test images. The project requires students to produce a mini scientific article/engineering report, a functional model with code, and a set of predictions submitted to the HuggingFace leaderboard.