CCPD Dataset 中国车牌检测数据集

CCPD 数据集全称为 Chinese City Parking Dataset，是一个大型的、多样化的、经过仔细标注的车牌识别数据集，由中科大团队于 2018 年建立，相关论文成果「Towards End-to-End License Plate Detection and Recognition: A Large Dataset and Baseline」已被 ECCV 2018 接受。这个数据集主要在中国合肥市的停车场采集得到，包含了多种复杂环境下的车牌照片，例如模糊、倾斜、阴雨天和雪天等，使得数据集在车牌识别任务中更具挑战性。该分为两个部分，CCPD 2019 和 CCPD 2020（也称为 CCPD-Green），其中 CCPD 2019 包含普通车牌（蓝色车牌），而 CCPD 2020 则包含新能源车牌（绿色车牌）。

CCPD dataset is fully named Chinese City Parking Dataset. It is a large-scale, diverse and meticulously annotated license plate recognition dataset, developed by the team from the University of Science and Technology of China in 2018. The associated paper titled "Towards End-to-End License Plate Detection and Recognition: A Large Dataset and Baseline" was accepted by ECCV 2018. This dataset is primarily collected from parking lots in Hefei, China, and contains license plate images under various complex scenarios, such as blurry, tilted, rainy and snowy weather, which renders it more challenging for license plate recognition tasks. It is divided into two subsets: CCPD 2019 and CCPD 2020 (also referred to as CCPD-Green). Specifically, CCPD 2019 includes standard blue license plates, while CCPD 2020 covers new energy green license plates.

搜集汇总

背景与挑战

背景概述

CCPD数据集是一个大型、多样化的中国车牌检测数据集，由中科大团队于2018年创建，包含CCPD 2019（普通蓝色车牌）和CCPD 2020（新能源绿色车牌）两部分，总规模超过30万张图像，大小13.11 GB。该数据集采集自合肥市停车场，覆盖模糊、倾斜、阴雨天和雪天等多种复杂环境，旨在提升车牌识别任务的挑战性和实用性，适用于目标检测和车牌识别研究。