基于传递学习的图像分类命令行工具
nnClassifier的Python项目详细描述
#nnclassify分类 用于对图像进行分类的命令行工具
Toxicology – the branch of science concerned with the nature, effects, and detection of poisons – has traditionally relied on expensive mammalian studies. However, due to the large number of environmental toxins that need testing, less expensive, high throughput alternatives are required. Planaria – a small asexual flatworm – provide an inexpensive and scalable solution. The small size of planaria not only makes them inexpensive to maintain, but also easy to image with a high-resolution camera. Researchers can run many test simultaneously, record the planarias reaction, and use computer vision techniques to analyze the results in a cost-effect and timely manner. The results of some of these tests can be analyzed using traditional computer vision techniques; however, many tests involve classifying features of the planaria – such as their body shape or eyes – as normal or abnormal. Deep convolutional neural networks have achieved state of the art results on image classification tasks, but they require a large training set of images. Researchers do not have the time or resources to hand-label sufficient images for this technique to be effective. Using transfer-learning, we have developed a software platform for researchers to quickly classify images with near state-of-the-art performance and minimal training data.
欢迎加入QQ群-->： 979659372
