摘要
由一名新闻记者-机器人和机器学习的工作人员新闻编辑每日新闻-详细的机器人-机器人和自动化数据已经呈现。根据NewsRx记者从法国亚眠发回的新闻报道，研究称，“作为林业领域许多机器人系统开发的一部分，森林场景理解需要使用计算机视觉算法。然而，这种密集和不受干扰的环境很复杂，需要对传统的检测方法进行测试。”这项研究的财政支持来自国家研究机构(ANR)。我们的新闻编辑从皮卡尔大学儒勒凡尔纳分校的研究中获得了一句话：“在树实例分割的情况下，存在密集间隔甚至相互交织的树，它们高度可变的形状，以及由于它们的枝叶而产生的复杂面具，这些都是我们面临的一些挑战。”为了更好地区分不同的树边界，我们提出了一种用于实时实例分割的卷积神经网络Consumal Mask，Consumal Mask选择了一种Labe L表示方法，该方法由树极值定义的凸多边形和二进制掩模来处理标签可能包含的细节和遮挡。
Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News – Data detailed on Robotics - Robotics a nd Automation have been presented. According to news reporting originating from Amiens, France, by NewsRx correspondents, research stated, “As part of the devel opment of many robotic systems for the forestry sector, forest scene understandi ng requires the use of computer vision algorithms. However, this dense and unstr uctured environment is complex and puts conventional detection approaches to the test.” Financial support for this research came from Agence Nationale de la Recherche ( ANR). Our news editors obtained a quote from the research from the University of Picar die Jules Verne, “In the case of tree instance segmentation, the presence of clo sely spaced or even intertwined trees, their highly variable shapes, and complex masks due to their branches and leaves are just some of the challenges to be ov ercome. For this, specific learning of tree boundaries is required to better dis tinguish one from another. In this letter, we propose ConvexMask, a convolutiona l neural network for real-time instance segmentation. ConvexMask opts for a labe l representation approach with a convex exterior polygon, defined by tree extrem ities, and a binary mask to handle the detail and occlusions that the label may contain.”
NETL
NSTL