首页
学习
活动
专区
工具
TVP
发布
精选内容/技术社群/优惠产品,尽在小程序
立即前往
您找到你想要的搜索结果了吗?
是的
没有找到

tf.image.non_max_suppression

贪婪地选择按得分降序排列的边界框子集。删除与先前选择的框具有高交叉-过度联合(IOU)重叠的框。边界框以[y1, x1, y2, x2]的形式提供,其中(y1, x1)和(y2, x2)为任意对角对角框角的坐标,坐标可以标准化(即,位于区间[0,1]或绝对区间。注意,这个算法不知道原点在坐标系中的什么位置。注意,这个算法对于坐标系的正交变换和平移是不变的;因此,坐标系统的平移或反射会导致算法选择相同的框。这个操作的输出是一组整数,索引到表示所选框的边界框的输入集合中。然后使用tf可以获得与所选索引对应的边界框坐标。收集操作。例如:selected_indices = tf.image。non_max_suppression(box, scores, max_output_size, iou_threshold)选择ted_boxes = tf。收集(盒、selected_indices)

02

[医疗信息化][DICOM教程]开篇介绍,新冠肺炎为医疗保健信息产业带来新的的紧迫性

The ongoing COVID-19 pandemic and the tragedies that have occured (and still occuring) have helped highlight the need for more timely exchange of critical healthcare related information for governments, health agencies, care providers and patients around the world. For many decades, the healthcare community has been at the forefront of standardization efforts for information exchange through the use of communication protocols such as HL7 and DICOM, and has worked hard to promote the use of these standards worldwide. However, the recent experience only highlights the fact that more opportunities exist to help achieve even more synergies and efficiencies in the information exchange processes that need to occur between various systems involved in the overall process of planning, administering, receiving and monitoring of all healthcare-related activities that are operationalized at any moment.

02
领券