MS COCO数据集多实例注释规则
[英]MS COCO dataset multiple instance annotation rule
问题描述
I am working with Mask-RCNN and want to train my own dataset with few categories of MS COCO dataset as well. 
我正在使用Mask-RCNN，并且想用几类MS COCO数据集来训练自己的数据集。 From MS COCO dataset I want to use Person, Bus, Car, Bicycle objects. 从MS COCO数据集中，我想使用“人”，“公共汽车”，“汽车”，“自行车”对象。
In my own dataset and I have annotated the images. 在我自己的数据集中，我为图像添加了注释。 My groundtruth is an image of same size and for every pixel I have a number which is the class ID. 我的事实是一张大小相同的图像，对于每个像素，我都有一个数字，即类ID。
Like for Person class my ground truth image has pixel colour (1,1,1) same as COCO dataset. 像Person类一样，我的地面真实图像的像素颜色（1,1,1）与COCO数据集相同。 My question is if there are two person in an image should both be annotated with colour (1,1,1) or is there a different rule? 我的问题是，图像中是否有两个人都应使用颜色（1,1,1）进行注释，或者是否存在其他规则？ As to show two instances of an object we need some kind of distinction. 为了显示一个对象的两个实例，我们需要某种区别。 Does anyone know how they are being annotated in the MS COCO dataset? 有谁知道在MS COCO数据集中如何对其进行注释？
1 个解决方案
解决方案1
0 2018-02-19 10:03:59
In case someone else is also facing the same issue: 如果其他人也面临相同的问题：
https://github.com/cocodataset/cocoapi/issues/111 https://github.com/cocodataset/cocoapi/issues/111
Check comments of me and nightrome. 查看我和nightrome的评论。
声明:本站的技术帖子网页，遵循CC BY-SA 4.0协议，如果您需要转载，请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.