PlantWild
收藏资源简介:
PlantWild是由昆士兰大学创建的大规模多模态植物疾病识别数据集,包含18,542张野生环境下的植物图像和89种疾病类型。数据集的图像来自多样化的互联网资源,每种疾病类别配有详细的文本描述,这些描述由GPT-3.5生成,旨在提供丰富的文本信息以辅助疾病识别。数据集的创建过程包括图像的收集、筛选和标注,确保了数据的质量和准确性。PlantWild数据集主要用于评估和改进在真实世界环境中植物疾病识别的方法,特别是在处理小类间差异和大类内变异问题时。
PlantWild is a large-scale multimodal plant disease recognition dataset developed by the University of Queensland, which contains 18,542 plant images captured in wild environments and covers 89 disease categories. The images of the dataset are sourced from diverse internet resources, and each disease category is provided with detailed textual descriptions generated by GPT-3.5, aiming to offer rich textual information to assist plant disease recognition. The dataset creation process includes image collection, filtering and annotation, ensuring the quality and accuracy of the data. The PlantWild dataset is mainly used to evaluate and improve plant disease recognition methods in real-world scenarios, especially when addressing the challenges of inter-class differences among minor categories and intra-class variations within major categories.

- 1Benchmarking In-the-wild Multimodal Disease Recognition and A Versatile Baseline昆士兰大学 · 2024年



