2020年宁夏枸杞虫害图文跨模态检索数据集

现存农业数据集缺乏枸杞虫害数据，且多为单一模态，所以构建一个可供跨模态识别研究的枸杞虫害数据集对于实现枸杞虫害识别防治具有重要意义。为此我们构建了枸杞虫害图文跨模态检索数据集，覆盖了枸杞虫害的图像采集、文本撰写、数据增强、分类、图文对应等多方面内容。本数据集目前共包含17类常见枸杞病虫害图像与文本描述数据。

Existing agricultural datasets lack data related to wolfberry pests and diseases, and most of them are single-modal. Therefore, constructing a wolfberry pest dataset for cross-modal recognition research is of great significance for realizing the identification and control of wolfberry pests and diseases.
To this end, we have developed a cross-modal text-image retrieval dataset for wolfberry pests and diseases, which covers multiple aspects including image collection, text writing, data augmentation, classification, and image-text correspondence.
This dataset currently includes image and text description data for 17 common types of wolfberry pests and diseases.