Hyperspectral Benchmark
收藏资源简介:
Hyperspectral Benchmark是一个综合性的数据集,由蒂宾根大学的认知系统团队创建,旨在解决高光谱图像(HSI)应用中的数据集限制问题。该数据集包含三个不同的HSI应用数据集:食品检测、遥感和回收。这些数据集的总大小约为250GB,可通过提供的下载脚本轻松访问。此外,还提供了一个PyTorch框架,用于复现结果和简单实现额外的模型。数据集的多样性支持了HSI模型的预训练管道,增强了训练过程的稳定性,并提供了一个处理小目标数据集大小的程序框架。
Hyperspectral Benchmark is a comprehensive dataset created by the Cognitive Systems Group at the University of Tübingen, designed to address the limitations of existing datasets in hyperspectral image (HSI) applications. This dataset encompasses three distinct HSI application datasets: food detection, remote sensing, and recycling. With an overall size of approximately 250 GB, it can be easily accessed via the provided download script. In addition, a PyTorch framework is provided for reproducing experimental results and readily implementing additional models. The diversity of this dataset supports the pre-training pipeline for HSI models, enhances the stability of the training process, and offers a procedural framework for handling small-scale target datasets.

- 1Hyperspectral Benchmark: Bridging the Gap between HSI Applications through Comprehensive Dataset and Pretraining认知系统,蒂宾根大学 · 2023年



