Farming Tool Usage Action Recognition Dataset

The current agricultural industry faces challenges such as labor shortages and low production efficiency. Traditional methods for monitoring farm tool usage often rely on manual recording, which is inefficient and prone to errors. Existing technical solutions have limitations in terms of real-time application and accuracy. Therefore, this dataset aims to support the smart development of agricultural automation equipment by providing high-quality action recognition data for farm tools. The dataset contains 5000 images of farm tool usage scenarios, captured with professional cameras in natural environments to ensure an authentic reflection of farming processes. We implemented multiple rounds of labeling and expert review to ensure the labeling accuracy exceeds 95%. All data is stored in JPG format and organized by category and action for quick and easy retrieval and use by users. The core advantage of this dataset is its high-quality labeling, which ensures consistency and completeness. Additionally, new data augmentation techniques enhance the diversity of the data, addressing the issue of insufficient samples in traditional datasets. Through this dataset, the accuracy of farm tool action recognition is expected to improve by 20%, significantly boosting agricultural production efficiency.