Nexdata | Global Vehicle Detection and Tracking Images | 3 Million Image | Computer Vision Data|Traffic Data for AI & ML Training | Identity Data

1. Specifications Data size : 3 million images Collecting environment : underground parking lot, surface parking lot, entrance and exit gates, outdoor roads (highways, urban roads, etc.) Data diversity : different surveillance scenes, different time periods, different cameras, various vehicle distributions (crowded, sparse) Device : surveillance camera, cellphone (a few) Collecting angle : looking down angle, eye-level angle Collecting time : day, night Data format : the image data format is .jpg, the annotation file format is .json Annotation content : vehicles rectangular bounding boxes and vehicle type attributes were annotated Accuracy : the bounding box of vehicle is qualified when the deviation is not more than 3 pixels, and the qualified rate of the bounding box shall not be lower than 97% 2. About Nexdata Nexdata owns off-the-shelf 200,000 hours of speech recognition data, 800TB of Annotated Imagery Data, about 2 billion pieces of Natural Language Processing (NLP) Data. These ready-to-go Identity Data supports instant delivery, quickly improve the accuracy of AI models. For more details, please visit us at https://www.nexdata.ai/computerVisionTraining?source=Datarade

搜集汇总

背景与挑战

背景概述

该数据集包含300万张车辆检测与追踪图像，采集于地下停车场、路面停车场、出入口及户外道路等多种监控场景，涵盖不同时段、摄像头角度和车辆密度。图像以.jpg格式提供，标注文件为.json格式，包含车辆边界框和类型属性，标注精度要求偏差不超过3像素且合格率不低于97%，适用于计算机视觉和交通AI模型训练。