Handheld Mercury Thermometer Hand Key Point Detection Dataset

As people place more importance on health monitoring, handheld mercury thermometers become commonly used temperature measuring tools, and their frequency of use is gradually increasing. However, current research on hand key point detection is limited, with existing datasets mostly concentrating on general object detection and lacking annotations for specific application scenarios. Additionally, existing solutions have shortcomings in annotation accuracy and consistency, failing to meet the needs for precise detection. This dataset aims to provide high-quality hand key point detection data to support research and applications of relevant algorithms. Data collection involves using high-resolution cameras under different lighting and backgrounds to ensure diversity and representativeness. Multi-round annotation and expert review methods are employed for quality control, ensuring annotation consistency. The data is stored in JPEG format, structured into fields including image ID, key point coordinates, bounding boxes, and labels, facilitating subsequent analysis and use. The core advantage of this dataset is its annotation accuracy, exceeding 95%, with consistency reaching 90%, enhancing the accuracy of key point detection. Additionally, employing new data augmentation techniques can effectively increase training samples, further improving model performance. Through practical application testing, this dataset helps to increase the recognition rate of related models by 15%, effectively addressing the inadequacy of traditional methods to adapt to complex scenarios.