停车场车位相机识别率分析数据

停车场车位检测相机广泛应用于停车场的车位信息检测，实现停车场停车收费、停车引导以及违停告警等场景功能。通过搭载高精度车位检测与车牌识别算法，可实现对地下停车场及地上停车楼等复杂光线场景的视频采集、车辆检测、车牌识别、数据传输、高清录像等功能。车位检测设备的识别率是车位识别相机设备性能评价的关键性指标，关系到车位状态的真实反馈。同时，设备应能够满足室内外复杂光线状况下保持高准确率，为停车引导及反向寻车等应用场景提供可靠的技术保障。1、数据采集：设备安装完成，分别对不同场景各进行200次测试，记录车位状态在室内昏暗光线场景、室内亮灯光线场景、室外明亮光线场景、室外日光直射光线场景这四种情况下识别结果；
2、数据计算：采用COUNTIFS公式计算正确次数；xx相对识别率（室内昏暗光线场景、室内亮灯光线场景、室外明亮光线场景、室外日光直射光线场景）=xx正确次数/200×100%；采用加权算法计算平均相对识别率，平均相对识别率=室内昏暗光线场景相对识别率*0.3+室内亮灯光线场景相对识别率*0.4+室外明亮光线场景相对识别率*0.2+室外日光直射光线场景相对识别率*0.1。
3、数据结论：统计平均相对识别率不低于99%，则判定为合格，反之则为不合格。

Parking space detection cameras are widely deployed for parking space information monitoring in parking facilities, supporting core functions including parking fee collection, automated parking guidance, and illegal parking alert across various parking scenarios. Integrated with high-precision parking space detection and license plate recognition algorithms, these devices can perform video capture, vehicle detection, license plate recognition, data transmission, and high-definition video recording in complex lighting environments, such as underground parking lots and above-ground parking garages.

The recognition accuracy of parking space detection devices is a critical performance metric for evaluating parking space recognition camera systems, as it directly determines the fidelity of parking space status feedback. Additionally, the devices must maintain high recognition accuracy under both indoor and outdoor complex lighting conditions, providing reliable technical support for applications such as parking guidance and reverse vehicle finding.

1. Data Collection: Upon successful installation, 200 tests are conducted for each of the four predefined lighting scenarios, and the recognition results of parking space status are recorded under the following four conditions: indoor dim lighting, indoor bright lighting, outdoor bright daylight, and outdoor direct sunlight exposure.
2. Data Calculation: The COUNTIFS formula is utilized to calculate the number of correct recognition results. The relative recognition rate for each scenario is calculated as: Relative Recognition Rate = (Number of Correct Recognitions in the Scenario / 200) × 100%, covering the four scenarios of indoor dim lighting, indoor bright lighting, outdoor bright daylight, and outdoor direct sunlight exposure. The weighted average relative recognition rate is computed using a weighted averaging algorithm, with the formula as follows:
Weighted Average Relative Recognition Rate = (Relative Recognition Rate of Indoor Dim Lighting × 0.3) + (Relative Recognition Rate of Indoor Bright Lighting × 0.4) + (Relative Recognition Rate of Outdoor Bright Daylight × 0.2) + (Relative Recognition Rate of Outdoor Direct Sunlight Exposure × 0.1)
3. Data Conclusion: The device is deemed qualified if the calculated weighted average relative recognition rate is no less than 99%; otherwise, it is classified as unqualified.

搜集汇总

数据集介绍

特点

该数据集包含800条停车场车位相机的识别率数据，记录了四种不同光线场景下的识别正确次数和相对识别率，用于评估车位检测设备的性能。数据每季度更新，主要应用于停车场的车位信息检测和管理场景。