手机室内视觉定位要素识别样本库MIPNS-LIB 3.0，是一个面向室内多场景条件下利用智能手机进行视觉定位的目标检测模型训练样本库。MIPNS-LIB 3.0在本团队此前提出的2.0版本中扩展了样本类别和样本数量。它包括建筑物室内场景中普遍出现的17类普适性的可识别视觉要素。本数据集是利用手机后置摄像头人工拍摄各视觉要素的样本照片，并对目标按照不同的角度和距离进行采集，通过LableImg标注工具完成制作。可适用于面向室内场景要素识别的目标检测模型的训练和验证。

MIPNS-LIB 3.0样本集在2.0版本的基础之上增加了样本数量并扩展了6类视觉要素，其中所包含的普适性地图定位锚点要素类别具体包括：消防栓、火警报警器、门、摄像头、WLAN、安全出口、电箱、门牌、电梯、展板、灯、书架、借书机、导向牌、充电桩、地铁闸机、到站显示屏共17类，可适用于办公楼、图书馆、商场、地铁站等不同建筑物室内场景要素识别任务。

MIPNS-LIB 3.0样本库共计包括27895个要素样本，每一类要素的比例统计如下表所示。

The Mobile Indoor Visual Positioning Element Recognition Sample Library MIPNS-LIB 3.0 is a sample library for training object detection models for visual positioning using smartphones in indoor multi-scenario conditions. MIPNS-LIB 3.0 expands on the sample categories and sample quantities from the previously proposed 2.0 version by our team. It includes 17 universally recognizable visual elements commonly found in indoor building scenes. This dataset was created by manually capturing sample photos of each visual element using the rear-facing camera of a smartphone, and collecting the targets at different angles and distances. The samples were annotated using the LabelImg annotation tool. It is suitable for training and validating object detection models for indoor scene element recognition.

The MIPNS-LIB 3.0 dataset builds upon the 2.0 version by increasing the number of samples and expanding six categories of visual elements. The specific categories of universal map positioning anchor points included are: fire hydrants, fire alarms, doors, cameras, WLAN, emergency exits, electrical boxes, door signs, elevators, bulletin boards, lights, bookcases, book lending machines, directional signs, charging stations, subway turnstiles, and arrival display screens, totaling 17 categories. This dataset is suitable for indoor scene element recognition tasks in various building types such as office buildings, libraries, shopping malls, and subway stations.

The MIPNS-LIB 3.0 sample library contains a total of 27,895 element samples. The proportion of each type of element is shown in the table below.