【摘要】 蝴蝶物种的自动识别能够为生态环境监测、生物多样性保护、害虫防治等领域提供技术支持,减少对专家人工鉴定的依赖。针对目前蝴蝶自动识别研究中主要依靠标本照片,缺少蝴蝶自然生态照片数据的状况,本数据集提供了一组蝴蝶生态照片、对应的PASCAL VOC 2007格式的标注文件以及每张照片的掩模。本数据集包含的721张中国蝴蝶生态图片,涵盖94种蝴蝶。每张照片对应一个标注文件,标注文件中包含照片大小、照片中蝴蝶的种类和位置等信息。每张照片的掩模对应照片中的蝴蝶区域。本数据集可为自动识别、目标检测、图像分割和昆虫分类等研究提供基础数据。更多还原

【Abstract】 While reducing reliance on experts, automatic species identification can provide strong technical assistance to ecological environment monitoring, biodiversity protection, pest control. However, all the datasets used for butterfly species auto-identification are currently based on specimen images, and there is a lack of ecological butterfly images. Therefore, we set up a dataset of butterfly images which includes ecological images of the butterfly species in China, corresponding annotation files in PASCAL VOC 2007 format, and a mask for each image. The dataset contains 721 butterfly images, covering 94 butterfly species. Each annotation file corresponds to an image, which contains the image size, taxonomic information and location of the butterfly. The mask file for each image corresponds to the specific location of the butterfly in the image. This dataset can support researches on automatic butterfly species recognition, object detection, image segmentation and insect taxonomy.更多还原