围栏检测

针对加油站卸油区场景，检测卸油期间，是否正确放置围栏（雪糕筒或黄黑色、银色等金属围栏），输出围栏和金属围栏标签，其中围栏指雪糕筒类型，金属围栏为黄黑色或银色等颜色。
精度对应场景或者限制条件
1）场景限定：加油站卸油区，油罐车在监控画面中；
2）图片分辨率：1080p
3）设备/镜头：支持RGB可见光识别；不支持红外识别
4) 识别对象尺寸：围栏高度不小于80px，遮挡不能超过1/2
5）识别的种类：放置于油罐车周围的安全围栏--（雪糕筒）

This dataset targets the oil unloading zone scenario of gas stations, aiming to detect whether safety barriers (traffic cones or yellow-black/silver metal fences) are properly placed during oil unloading operations, and output the labels for barriers and metal fences, where "barriers" specifically refer to traffic cone-type barriers, while "metal fences" denote those in colors such as yellow-black or silver.

Constraints and scenario specifications corresponding to detection accuracy are listed below:
1. Scenario limitation: Oil unloading zone of gas stations, with the oil tanker visible in the monitoring footage;
2. Image resolution: 1080p;
3. Device/lens requirements: Supports RGB visible light recognition only, and does not support infrared recognition;
4. Recognized object size requirements: The height of the fence shall be no less than 80 pixels, and the occlusion shall not exceed 1/2;
5. Recognized object categories: Safety barriers placed around the oil tanker -- (traffic cones)