Manual Spraying Action Recognition Dataset

The current agricultural sector faces challenges such as low spraying efficiency and environmental pollution caused by improper use of pesticides. Existing solutions often rely on manual monitoring, which is inefficient and prone to errors. This dataset aims to help developers create computer vision-based automated recognition systems by providing high-quality manual spraying action data, thereby enhancing the automation and intelligence level of spraying operations. Data collection is performed using high-resolution cameras in real agricultural environments to ensure data authenticity and diversity. We conducted multiple rounds of annotation and consistency checks, reviewed by professionals to ensure high data quality and reliability. Data is stored in JPG format with corresponding metadata for each image, facilitating subsequent processing and analysis. The core advantage of this dataset lies in its high annotation accuracy and integrity, with annotation consistency reaching over 95%, significantly surpassing the industry average. By introducing new data augmentation techniques, the model's performance in complex environments is effectively improved. In practical applications, models trained using this dataset have seen a 15% increase in spraying action recognition accuracy, greatly improving pesticide use efficiency and reducing environmental pollution risks.