Nematode worms are one of most abundant metazoan groups on the earth, occupying diverse ecological niches. Accurate recognition or identification of nematodes are of great importance for pest control, soil ecology, bio-geography, habitat conservation and against climate changes. Computer vision and image processing have witnessed a few successes in species recognition of nematodes; however, it is still in great demand. In this paper, we identify two main bottlenecks: (1) the lack of a publicly available imaging dataset for diverse species of nematodes (especially the species only found in natural environment) which requires considerable human resources in field work and experts in taxonomy, and (2) the lack of a standard benchmark of state-of-the-art deep learning techniques on this dataset which demands the discipline background in computer science. With these in mind, we propose an image dataset consisting of diverse nematodes (both laboratory cultured and naturally isolated), which, to our knowledge, is the first time in the community. We further set up a species recognition benchmark by employing state-of-the-art deep learning networks on this dataset. We discuss the experimental results, compare the recognition accuracy of different networks, and show the challenges of our dataset. We make our dataset publicly available at: https://github.com/xuequanlu/I-Nema
翻译:网球虫是地球上最富的元动物群之一,占据着不同的生态位置。对线虫的准确认识或识别对于虫害控制、土壤生态、生物地球学、生境保护和应对气候变化非常重要。计算机视觉和图像处理在物种识别线虫方面取得了一些成功;然而,它仍然有很大的需求。在本文件中,我们确定了两个主要瓶颈:(1) 缺乏关于各种线虫(特别是仅在自然环境中发现的物种)的公开成像数据集,这需要大量实地人力资源和分类学专家;(2) 缺乏关于这一数据集的先进深层学习技术的标准基准,这需要计算机科学的学科背景。我们铭记着这一点,提出由多种线虫(实验室培养和自然孤立)组成的图像数据集,据我们所知,这是社区中的第一次。我们进一步设置了物种识别基准,在数据库中采用最新的深层学习网络。我们讨论了实验结果,对不同数据的准确度进行了比较。我们用不同数据网络/图表的精确度进行了对比。我们用不同数据来评估了我们的数据。