Mechanical-harvest Soybean Images Dataset
Description
Machine vision and deep learning technology are widely applied to detect the quality of mechanized soybean harvesting. A clean and tidy dataset is the foundation for constructing an online detection learning model for the quality of mechanized harvested soybeans. To achieve this goal, we created an image dataset for mechanized harvesting of soybeans. These photos were taken on October 9, 2018 at a soybean experimental field of Liangfeng Grain and Cotton Planting Professional Cooperative. This dataset contains 40 soybean images of different qualities. By scaling, rotating, flipping, filtering, and adding noise to enhance the data, the dataset was expanded to 800 frames. The dataset consists of three folders, which store images, label maps, and record files for partitioning training, validation, and testing sets. The author has confirmed the availability of this dataset through previous research. This dataset can help researchers construct a quality prediction model for mechanized harvested soybeans using deep learning techniques. Collected images of mechanized harvesting of soybeans. The industrial camera (1080P(V5610)_PCBA, Midway Vision Technology, Guangzhou, China) is used to capture images of soybeans harvested using a soybean combine harvester. The original image format is jpg, with a size of 1280 * 1024 pixels. These images of mechanically harvested soybeans with different qualities were taken on October 9, 2018 at the soybean testing platform of Liangfeng Grain and Cotton Planting Professional Cooperative in Guanyi District, Liangshan City, Shandong Province. The shooting process is supplemented with LED visual light source, with coordinates of 35.28278 ° N and 116.54047 ° E. A total of 40 images were captured. These images were enhanced by scaling, rotating, flipping, filtering, and adding noise, expanding to 800 images. Manually annotate images, annotate category and location information, and save annotations in PNG format for use by different models. The dataset is divided into training, validation, and testing sets suitable for machine learning. The folder size of the dataset is 328 MB and a RAR file is provided for easy download.