手套检测

针对厨房等要求手套佩戴的室内场景，进行手套检测
使用场景：厨房、食堂、冷冻仓库、市场等各类食品生产场所。
精度对应场景或者限制条件
图片分辨率：1080p及以上。
2）设备/镜头：支持RGB可见光彩色图像；不支持红外图像。
3）成像条件：画面清晰，无明显拖尾、抖动、异常颗粒、偏色等模糊情况。画面曝光合适，无过曝引起的高光溢出，无强逆光。目标边缘清晰，与背景对比鲜明，肉眼可分辨属性。
4) 算法边界：
遮挡：人体遮挡小于 30%，人手完全不能遮挡，人手五指需要清晰可见
单图人头数量小于10；
尺寸：以1920*1080分辨率图像为基准，人手目标大于30*30、人体60*120px，相应比例调整目

This dataset is designed for glove detection in indoor scenarios where glove wearing is mandatory, such as kitchens.
Application scenarios: various food production places including kitchens, canteens, cold storage, markets, etc.

Scenario and constraint requirements for detection accuracy:
1. Image resolution: 1080p or higher.
2. Equipment/lens: Supports RGB visible light color images; does not support infrared images.
3. Imaging conditions: The image must be clear, free of obvious smearing, jitter, abnormal artifacts, color cast and other blurring issues. The exposure shall be appropriate, with no highlight clipping caused by overexposure and no strong backlighting. The edges of the target are distinct, with high contrast against the background, and the target attributes can be distinguished by the naked eye.
4. Algorithm constraints:
- Occlusion: Human body occlusion rate less than 30%, no complete occlusion of human hands, and the five fingers of each hand must be clearly visible.
- Number of human heads per image: less than 10.
- Target size: Taking 1920*1080 resolution images as the benchmark, the human hand target should be larger than 30*30px, and the human body target should be larger than 60*120px, with corresponding proportional adjustment for the target size.

搜集汇总

背景与挑战

背景概述

该数据集是一个用于检测厨房等室内场景中手套佩戴情况的算法模型，适用于多种食品生产场所，并对图像质量和检测条件有具体要求。